Skip to main content

Full text of "Recommendations and Reports - 1982, Volume 2"

See other formats

^™^ice of \t 

fice of the Chairman 
Administrative Conference of the United States 


of the 
iJnited States 

nd Reports 



Administrative Conference of the United States 

The Administrative Conference of the United States was established by statute as an independent 
agency of the federal government in 1964. Its purpose is to promote improvements in the 
efficiency, adequacy and fairness of procedures by which federal agencies conduct regulatory 
programs, administer grants and benefits, and perform related governmental functions. 

To this end, the Conference conducts research and issues reports concerning various aspects of the 
administrative process and, when warranted, mokes recommendations to the President, Congress, 
particular departments and agencies, and the judiciary concerning the need for procedural 
reforms. Implementation of Conference recommendations may be accomplished through direct 
action on the part of the affected agencies or legislative changes. 


of the 
United States 


OCT 2 2 1984 

and Reports 


Volunne II 




Vol. Page 

Recommendations of the Administrative Conference of the 
United States; 

Recommendation 82-1: Exemption (b)(4) of the Freedom of In- 
formation Act 1 1 

Recommendation 82-2: Resolving Disputes Under Federal Grant 

Programs I 9 

Recommendation 82-3: Federal Venue Provisions Applicable to 

Suits Against the Government I 15 

Recommendation 82-4: Procedures for Negotiating Proposed 

Regulations 1 21 

Recommendation 82-5: Federal Regulation of Cancer-Causing 

Chemicals II 1 

Recommendation 82-6: Federal Officials' Liability for Constitu- 
tional Violations II 13 

Recommendation 82-7: Judicial Review of Rules in Enforcement 

Proceedings II 15 

Background Reports for Recommendations 82-1 
THROUGH 82-7: 

82-1 James T. O'Reilly. Regaining A Confidence: Protection 
of Business Confidential Data through Reform of the 

Freedom of Information Act I 29 

Russell B. Stevenson, Jr. Protecting Business Secrets Un- 
der The Freedom of Information Act: Managing 

Exemption 4 I 81 

82-2 Ann Steinberg. Federal Grant Dispute Resolution I 137 

82-3 Schlesinger, Steve R., Editor. Venue at the Crossroads I 219 
82-4 Philip J. Harter. Negotiating Regulations: A Cure for 

Malaise I 301 

82-5 Richard A. Merrill. Federal Regulation of Cancer Causing 

Chemicals II 21 

82-6 Thomas J. Madden & Nicholas W. Allard. Advice on 

Official Liability and Immunity II 20 1 

82-7 Paul R. Verkuil. Congressional Limitations on Judicial 

Review of Rules: The Jurisprudential Implications II 443 


Statement on Informal Rulemaking Provisions of the Administrative 

Procedure Act II 486 

Statement on Discipline of Attorneys Practicing Before Federal 

Agencies II 487 



Vol. Page 

Report of the Administrative Conference Committee on Govern- 
mental Processes Concerning Discipline of Attorneys Practicmg 
Before Federal Agencies II 488 

Michael P. Cox. Report on Regulation of Attorneys Practicing 

Before Federal Agencies II 49 1 


Teresa M. Schwanz. The Consumer Product Safety Commission: 

A Flawed Product of the Consumer Decade II 578 



(Adopted June 18, 1982) 

The following recommendations broadly address the procedures by which 
federal agencies identify, evaluate, and regulate substances that pose a potential 
risk of human cancer. For many years these regulatory activities have been 
among the most controversial engaged in by federal agencies. They implicate 
important health and economic values and attract the interest of groups throughout 
society. Some of the issues dealt with in these recommendations are not peculiar 
to the context of carcinogen regulation. The statutory procedures agencies follow 
in regulating human exposure to cancer hazards, for example, are applicable to 
many other environmental and health hazards. While these recommendations 
may thus have broader application or impact, they are based on an evaluation 
of agency performance in this context alone. 

I. Priority Setting 

Estimates of the number of chemicals that may pose a risk of human cancer 
describe a universe larger than government agencies can evaluate or regulate. 
Agency resources for scientific review and regulatory proceedings are shrinking, 
and the capacity of manufacturers and users of chemicals to implement costly 
controls is likewise limited. In these circumstances, attention should be concen- 
trated on those chemicals that pose the greatest risks and can be controlled most 
economically. As agencies have increasingly recognized, accomplishing this 
objective requires establishment of priorities. 

Priority setting should be part of any program directed at determining the 
health effects of chemicals, as well as of programs designed to establish exposure 
limits. Each regulatory agency should set its own priorities, but there is also 
value in interagency selection of candidates for regulation and, particularly, for 
further testing. 

Although candidates for evaluation and regulation should be carefully se- 
lected, any ranking based on abstract criteria, however rational, will be vulnerable 
to new information about human exposure or health effects and to public con- 
cerns. Agencies should explain departures from established priorities, but should 
retain flexibility to respond to new problems. 

Criteria for selecting candidates for regulation should include the extent of 
the hazard posed by a chemical — a function of its potency, the conditions of 
exposure, and the number of people exposed. The extent of the hazard can often 



be expressed in quantitiative terms. Agencies should also consider the effec- 
tiveness and cost of alternative control measures as well as other effects of 
regulation. Estimates of the cost-effectiveness of regulating specific chemicals 
can be helpful in selecting priorities. Because of the complexity of these criteria 
and the continuing need to obtain additional information, selection of candidates 
for regulation will often be an iterative process. 

Priority setting should be a public process which affords interested persons 
an opportunity to communicate their own views to the agency. Varied techniques 
exist short of rulemaking under the Administrative Procedure Act for providing 
such opportunities, including public meetings, use of expert advisory panels, 
and publication in the Federal Register of invitations to submit data and views. 
Under whatever system, preliminary rankings risk being misinterpreted as de- 
finitive judgments about the hazards of chemicals before manufacturers and users 
have had an opportunity to present evidence or arguments. Accordingly, agency 
announcements of priorities should make clear the tentative character of under- 
lying scientific assessments and describe the opportunities for interested persons 
to supply contrary or confirmatory information. 


1 . To the extent compatible with other demands on their resources, agencies 
should establish and follow systems for ranking chemicals as candidates for 
scientific evaluation or regulation. Agencies should generally adhere to the prior- 
ities thus established, but must retain the tlexibility to respond to new hazards 
or to public concerns about other chemicals. 

2. Priority setting should be part of selecting chemicals (a) for further 
testing, (b) for intensive scientific and regulatory evaluation, and (c) for admin- 
istrative action to limit or eliminate exposure. 

3. To the extent allowed by law, priorities should be set by agencies with 
the objective of maximizing the net benefits to society of agency action. Thus, 
in setting priorities, the agency should consider the expected benefits and costs 
of various alternatives. Although the criteria for selecting chemicals for regulation 
should take account of the health hazards posed by chemicals, including the 
potency of the chemicals, levels of exposure, the number of people likely to be 
exposed, and the costs and cost-effectiveness of methods for controlling expo- 
sure, to the extent that these may be known. 

4. Because agencies differ in their functions and their selection of candi- 
dates for evaluation and regulation must be coordinated with other agency pro- 
grams, each agency should establish its own priorities. The several agencies 
responsible for regulating carcinogens should, however, periodically compare 
their rankings and, where feasible, coordinate their testing, evaluation, and 
regulatory efforts. Consultation with scientific, industrial, and public interest 
organizations should be encouraged. 

5. Agency procedures for setting priorities should permit interested mem- 
bers of the public to submit information and views concerning specific candidates. 
Agencies should consider adopting priority-setting systems through rulemaking, 
but more informal methods are appropriate for ranking individual chemicals for 


evaluation and regulation. Any agency announcement that a product or chemical 
is a candidate for further evaluation or for regulation should explain both the 
bases of that assessment and its preliminary nature and should describe available 
procedures for confirming or negating its preliminary conclusions. 

II. Interagency Coordination 

Responsibility for regulating chemicals that pose a risk of human cancer is 
shared by several agencies, including most notably the Environmental Protection 
Agency (EPA), Occupational Safety and Health Administration (OSHA), Food 
and Drug Administration (FDA), and Consumer Product Safety Commission 
(CPSC). Only in recent years have these agencies attempted to coordinate their 
activities to assure consistency and enhance the benefits of regulation. While 
the agencies' criteria for identifying and evaluating potential carcinogens have 
in fact not been conflicting, their decisions have sometimes appeared difficult 
to reconcile, usually because of differences in legislation. 

The creation of the Interagency Regulatory Liaison Group in 1977 rested 
on the premises that federal regulation of chemical carcinogens should proceed 
from common principles of scientific evaluation and should be coordinated ad- 
ministratively. While these premises remain valid, other mechanisms may be 
equally effective in assuring coordination. Opportunities for productive coor- 
dination exist in several areas: establishment of government-wide principles of 
scientific evaluation; agreement on test guidelines for toxicological experiments 
and other studies of health effects; ranking of chemicals for testing; and moni- 
toring and enforcement of exposure controls. 

The four agencies have on several recent occasions cooperated in the reg- 
ulation of specific chemicals. They once assembled a central scientific working 
group to evaluate formaldehyde, a chemical of common interest, and they agreed 
on a common list of chemicals that were high priorities for coordinated regulation. 
Though diverse statutory procedures have sometimes discouraged joint admin- 
istrative proceedings, opportunities at the preproposal stage for cooperation in 
evaluating health effects and performing economic analysis have not yet been 
fully exploited. 


1. Interagency coordination in identifying, evaluating, and regulating po- 
tential human carcinogens should be encouraged. Effective coordination can 
reduce governmental costs, minimize inconsistency among the agencies, and 
better illuminate the economic costs of alternative control options. 

2. Agencies should continue to cooperate in identifying chemicals for which 
further testing is needed to permit regulatory assessments. The National Toxi- 
cology Program should continue to elicit joint agency rankings of candidates for 

3. Regulatory agencies should collaborate with government scientific bod- 
ies, including the National Toxicology Program, National Institute of Environ- 
mental Health Sciences, and National Center for Toxicological Research, to 
obtain agreement on guidelines for the conduct and evaluation of toxicological 


tests. This effort need not result in public rulemaking to adopt test protocols, 
but scientists outside government should have an opportunity to contribute to 
the development of test guidelines. Cooperative guidelines should also be con- 
sidered for short term tests and epidemiological studies, but current guidelines 
should not discourage development of improved test methods. 

4. To the extent permitted by statute, agencies responsible for regulating 
carcinogens should adhere to common criteria for evaluating and interpreting 
health effects data. Agencies should avoid inconsistency in their approaches to 
mixed scientific-policy issues, such as whether to assume a no-threshold model 
of carcinogenesis, whether to perform quantitative estimates of human risk, or 
whether to allow evidence that a chemical produces an increase in cancer in 
laboratory animals through mechanisms that do not suggest human risk. 

5. Agencies responsible for regulating carcinogens should continue to ex- 
plore joint evaluation of the potential health hazards of chemicals that are can- 
didates for regulation by more than one agency, for instance, by use of a multi- 
agency advisory panel for any particular substance. 

6. Agencies should explore other opportunities to collaborate before the 
initiation of formal administrative proceedings to regulate a chemical. These 
may include joint development of exposure estimates and joint preparation of 
economic analyses of alternative regulatory approaches. 

7. Agencies should, after eliciting the views of interested persons, consider 
conducting joint administrative proceedings when they contemplate regulating 
the same chemical. Statutory diversity may, however, complicate or even pre- 
clude such proceedings. 

HI. Chemical Selection and Guidelines for Testing and Evaluation 

The selection of chemicals for study, the design of protocols for laboratory 
experiments, and the establishment of criteria for interpreting study results are 
important parts of the process for regulating substances that pose a risk of human 
cancer. These are essentially scientific functions which should not be dictated 
by narrow policy considerations. The important desiderata are scientific integrity 
and consistency. 

At the same time, it remains true that much toxicological testing performed 
in the United States — by government laboratories and by or on behalf of private 
industry — is performed to aid regulatory decisionmaking. Numerous toxicolog- 
ical studies are performed to support marketing approval for food additives, 
pesticides, and pharmaceuticals. FDA and EPA have both conducted similar 
studies on substances within their regulatory jurisdiction. The agencies have also 
relied on studies commissioned by the National Cancer Institute. The NCI bioas- 
say program has more recently fallen under the supervision of the National 
Toxicology Program, whose supervisory board includes representatives of the 
regulatory agencies. NTP currently accepts nominations of chemicals for testing 
from the agencies. 

Under the auspices of the former Interagency Regulatory Liaison Group, 
the four agencies studied here (EPA, OSHA, FDA, and CPSC) actively sought 


agreement on joint protocols for toxicological testing and on criteria for evalu- 
ation of study results. 


1 . In conjunction with the National Toxicology Program and in consultation 
with scientific organizations outside government, regulatory agencies should 
continue efforts to develop consistent guidelines for toxicological testing. These 
test guidelines should reflect current scientific consensus but also awareness of 
the resource limitations that constrain both government and industry. The guide- 
lines developed or espoused by the agencies should not be issued as formal 
regulations. Rulemaking would needlessly prolong the process of reaching agree- 
ment and formal regulations would limit the ability of agencies, producers, and 
testing laboratories to design protocols or adapt existing guidelines to new cir- 

2. Agencies should adhere to similar test guidelines and similar criteria for 
interpreting test results. Any departure from common standards should be spe- 
cifically and convincingly justified. 

3. The National Toxicology Program should continue to encourage the 
participation of EPA, OSHA, FDA, and CPSC in its selection of chemicals for 
testing. It should be willing, on request, to assist agencies in their evaluation of 
study findings. The NTP special working group to evaluate the carcinogenicity 
of formaldehyde for all of the four regulatory agencies represents one promising 
experiment in the effort to assure consistency among the agencies. 

IV. Advisory Panels 

Assessment of the health risks posed by chemicals requires substantial 
information about their toxicity and about the conditions under which humans 
are exposed to them, as well as understanding of the techniques for obtaining 
such information. Techniques for toxicological evaluation, exposure estimation, 
and quantitative assessment of human risk increasingly demand state-of-the-art 
expertise. Evaluation of potential human carcinogens also requires impartiality 
in assessing the quality and interpreting the results of toxicological and epide- 
miological studies. In the scientific world, new research findings are customarily 
exposed to peer review before they become accepted as reliable. Increasingly, 
regulatory agencies have concluded that the data on which they rely, and the 
interpretations they give these data, should be subjected to peer review before 
becoming the bases for regulatory decisions. 

Several proposals have recently been advanced to mandate and institution- 
alize this peer review function. Some of these proposals call for strict separation 
of risk assessment from the process of evaluating and selecting regulatory options. 
These proposals raise difficult issues respecting the design of procedures for 
administrative decisionmaking about chemical carcinogens. The function of as- 
sessing risk can be distinguished analytically from the choice of regulatory 
responses, but separating them in practice is more difficult. The close relation- 
ships among issues of toxicity, exposure, and the cost of controls makes the 


process of agency evaluation an interactive one. Accordingly, mechanisms for 
exposing agency judgments to peer review should be flexible enough to permit 
frequent interchange with agency policy makers. 

One method that agencies have successfully used for institutionalizing peer 
review is consultation with expert advisory panels. These panels have taken 
several forms, ranging from informal working groups of agency scientists as- 
signed to review a single chemical to standing committees of independent experts 
who advise on numerous candidates for regulation. Advisory panels can con- 
tribute objectivity as well as expertise to agency decisions. Their advice has 
sometimes prevented erroneous regulatory actions; more frequently, their role 
has been to illuminate complex issues and enhance the quality, and thus the 
credibility, of agency scientific analysis. 

Advisory panels comprised in whole or in part of non-federal employees 
are governed by the Federal Advisory Committee Act (FACA), unless exempted 
by specific legislation. The FACA requires individual chartering and biennial 
reapproval and mandates practices that are designed to assure the balance, open- 
ness, and integrity of advisory committees. Most of these requirements are 
salutary, and the following recommendations endorse their observance by panels 
even where not formally subject to the FACA. Certain of the Act's provisions, 
however, have discouraged agencies from seeking outside peer review. The 
recommendations advanced here are intended to encourage agency resort to 
expert advisory panels without necessarily endorsing every requirement of the 
FACA. None of these recommendations should be construed as superseding 
either that Act, where applicable, or federal conflict of interest laws. 


1 . Peer review of experimental findings and scientific judgments is an 
important means of validating the technical bases of regulatory decisions con- 
cerning carcinogens. To the extent compatible with existing law, agencies should 
structure their decisional processes to incorporate mechanisms for scientific peer 

2. Expert advisory panels represent one valuable means for obtaining sci- 
entific peer review of agency decisions. Advisory panels can provide information 
that will aid agencies in setting priorities, in evaluating scientific data prior to 
initiation of administrative proceedings, and in evaluating evidence submitted 
by interested persons during public proceedings. 

3. The design, composition, and operation of an advisory panel should fit 
the function it is to perform; no uniform approach is optimal. A standing advisory 
panel, with responsibility for reviewing the scientific bases of major actions by 
a particular program, can contribute consistency in addition to expertise, and it 
provides a ready forum for agency consultation. To assure that the panel has 
access to relevant expertise, sub-units can be appointed to evaluate specific 
chemicals or issues. An agency that lacks a standing advisory panel, however, 
should not forgo opportunities to create ad hoc panels to review the scientific 
bases of contemplated regulatory actions. 


4. The role of an expert advisory panel may embrace evaluating data con- 
cerning the health effects of chemicals, interpreting those data and characterizing 
the chemicals' effects, and estimating the likely frequency of those effects under 
different exposure conditions. When an agency rejects an advisory panel's sci- 
entific judgment, it should explain the bases for that rejection. When an agency 
selects a regulatory approach whose bases appear inconsistent with a panel's 
advice, it should explain the legal, social, or other reasons that dictate or justify 
that choice. 

5. Members of an expert advisory panel should be selected primarily for 
their expertise in relevant scientific fields. Qualified scientists, even if employed 
by the agency they are to advise, by other government agencies, or by commercial 
organizations, may appropriately be selected to serve on advisory panels. In the 
selection of panel members, attention should be given to assuring balance in 
scientific orientation and viewpoint. The organizational affiliations of all panel 
members should be a matter of public record. The financial holdings and rela- 
tionships of potential panel members should be carefully screened by the ap- 
pointing officials and periodically reviewed thereafter to prevent conflicts of 
interest. If a panel includes members who are not disinterested with respect to 
a particular substance, those members should not participate in the panel's dis- 
cussions of that substance. 

6. Advisory panels should be accessible for consultation by the appointing 
agency at frequent intervals and on short notice. Consultation should usually 
occur before any announcement of a plan to regulate a chemical. Even when 
the Federal Advisory Committee Act does not apply, many of its requirements 
represent appropriate guides for the operation of advisory panels. Referrals of 
issues to advisory panels and meetings of agency officials with panel members 
should be matters of public record. Advance public notice of panel meetings 
should be provided where practicable. Panels should meet in open session except 
when reviewing data that are entitled to confidential treatment or, unless restricted 
by statute, when the panel members vote to close their deliberations. All panel 
conclusions and recommendations should be reduced to writing and become a 
part of the material to be considered in any ensuing administrative proceeding. 

7. Advisory panels established to evaluate scientific data for, and provide 
advice to, more than one agency may often be useful. Such panels can be 
particularly valuable in recommending chemicals for further study or testing. 
Where a chemical is of interest to more than one regulatory agency, a single 
advisory panel may be an efficient way of obtaining an independent assessment 
of its potential health effects. Interagency panels should be subject to the same 
restrictions as to composition, operation, and scope of responsibility as panels 
appointed to serve a single agency. 

V. Generic Rulemaking 

Agencies responsible for regulating potential carcinogens have attempted 
to develop and publish criteria for evaluating scientific data that underlie their 
decisions. These efforts have included informal statements of policy, proposed 
interpretative regulations, binding substantive standards, and interagency policy 


Statements. The motives underlying these efforts to estabhsh a framework for 
evaluating evidence concerning individual chemicals have been as diverse as the 
forms they have taken. One objective has been to obtain agreement among the 
agencies on the scientific criteria and policies that inform regulation of carcin- 
ogens. Another objective has been to improve understanding, among the public 
and within agency staffs, of the principles that guide agency decisions. Some 
agencies have also attempted to frame these principles as rules of decision in 
Order to forestall repetitive disputes in proceedings to regulate individual chem- 

The desirability of assuring agency adherence to common principles of 
scientific interpretation seems clear. Equally important is conformity of these 
principles with the best current understanding of the mechanisms of carcino- 
genesis, of toxicological and epidemiological research, and of qunatitative risk 
assessment. Attempts to treat decisional guides as though they were binding 
substantive rules may conflict with the need to remain sensitive to developments 
in rapidly changing fields. 

The efforts of individual agencies to establish criteria for identifying and 
evaluating potential human carcinogens have been criticized on several counts. 
In addition to disputing the content of agency judgments, industry critics have 
questioned the appropriateness of analytic frameworks that discourage indivi- 
dualized assessment of the capacity of chemicals to cause cancer. The duration 
of agency rulemaking proceedings suggests difficulty in obtaining scientific 
agreement in this area, but the controversy more often betrays disagreement over 
agency policy judgments rather than over their distillation of scientific consensus. 


1. In appropriate cases generic rules legitimately may be the basis for 
summary administrative resolution of recurrent issues, provided they do not 
foreclose reexamination of scientific conclusions respecting carcinogenicity of 
particular substances. Agencies should proceed cautiously in using this technique 
because the complexity and uncertainty of the issues involved and continuing 
advances in scientific understanding of the mechanisms of human cancer impede 
development of binding general principles. 

2. Agencies should be encouraged to develop systematic statements of the 
principles that they will apply in identifying, ranking, and evaluating chemicals 
that may pose a risk of human cancer. The systematizing process should ordinarily 
involve opportunity for submission of data and views by interested persons 
outside the agency. Whenever an agency expects to limit argument over principles 
that will guide decisions in proceedings to regulate individual chemicals, com- 
pliance with statutory requirements for rulemaking is essential before the prin- 
ciples are formulated. 

3. Agency statements may appropriately address the design and interpre- 
tation of scientific studies, the measurement or estimation of human exposure, 
the performance of quantitative risk assessment, and the selection of regulatory 
responses. These statements should attempt to distinguish between elements that 
are intended to summarize current scientific consensus and others that represent 


policy judgments reached in the absence of consensus. PoHcy judgments are an 
inevitable component of regulatory decisions and, where similar issues arise 
recurrently, are appropriately resolved on a generic basis. But where scientific 
developments in the near term are likely to require modification, or where 
individual studies or chemicals are often likely to deviate from the "norm," 
they should not be framed as binding rules. 

4. Many issues involved in identifying, evaluating and regulating potential 
carcinogens are common to all agencies and should be resolved consistently. 
These issues range from the criteria for interpreting scientific studies to the 
selection of mathematical models for estimating human risk. Though scientific 
uncertainty surrounding many of these issues requires considerations of value 
and policy, disparate agency responses should be avoided or convincingly jus- 

VI. Quantitative Assessment of Risk 

Various pressures and incentives have encouraged regulatory agencies to 
explore methods for quantifying the risk of potential carcinogens. At the same 
time the use of quantitative risk assessment in making regulatory decisions has 
provoked fierce controversy. Industrial interests have generally urged agencies 
to regard quantification as an essential step in evaluating measures for controlling 
carcinogens. Consumer groups and labor unions, on the other hand, have cau- 
tioned against excessive reliance on techniques whose reliability remains un- 
certain. The agencies themselves have for some time had difficulty reaching 
consensus on the issue. 

Criticism of quantitative risk assessment stems in part from doubts about 
its reliability. Reliance on animal models as qualitative predictors of human risk 
often may be unavoidable, and the further extrapolation from effects observed 
at high doses to the unmeasurable effects predicted at low doses compounds 
uncertainty. Critics also point to the wide range of risks at low doses predicted 
by different extrapolation models. A further criticism is that quantitative risk 
assessment can too easily be exploited to compute the dollar value implicitly 
assigned by regulators to human life. 

Despite these criticisms, quantitative risk estimation has an appeal for both 
analysts and decisionmakers. Without some means of describing the magnitude 
of the health effects associated with exposure to a carcinogen an agency must 
find some other basis for deciding what controls to require. Legislation sometimes 
provides an answer. The Delaney Clause, for example, makes quantification of 
the risk of a carcinogenic food additive superfluous. Under any statute that 
permits or instructs the agency to weigh the costs against the health consequences 
of alternative means of controlling exposure, however, a method to quantify 
risks has proved essential. 

Quantitative risk assessment can help illuminate many of the choices that 
agencies confront in regulating carcinogens. Even with its uncertainties, the 
technique facilitates comparison of the risks posed by different substances, which 
can aid in establishing priorities for regulation. Similarly, quantitative risk as- 


sessment can illuminate the choice among diverse regulatory options. A common 
unit of measurement for evaluating the health benefits of different options can 
materially advance analysis of a multidimensional decision even if the mea- 
surements are unverifiable and the benefits are not converted into dollars. 


1 . Quantitative risk estimates can be valuable in setting priorities for reg- 
ulation of carcinogens, comparing the human health consequences of alternative 
control measures, and in analyzing the costs and benefits of regulatory options. 
To the extent regulatory statutes allow and available data permit, agencies should 
attempt to estimate and describe the magnitude of the risk posed by prevailing 
levels of exposure to substances considered for regulation. Within the same 
constraints, agencies should also attempt to describe the size of the health benefits 
provided by measures required to reduce or eliminate human exposure. 

2. Given the limitations of techniques for quantitative risk assessment and 
different statutory criteria for limiting or eliminating exposure, risk estimates 
ordinarily will be only one consideration. The weight accorded such estimates 
should reflect: 

(a) the statutory criteria governing agency decisions; 

(b) the adequacy of available data on carcinogenic potency and on the type, 
levels, and duration of human exposure; and 

(c) the acceptance of the methods used to estimate future health effects. 

3. Any description of the magnitude of the risk associated with prevailing 
exposures or of the estimated health benefits of exposure controls should ex- 
plicitly identify: 

(a) the toxicological, epidemiological, and exposure data on which it rests; 

(b) the assumptions underlying any extrapolations from animals to man or 
from high to low exposure levels; 

(c) other assumptions about the behavior of the substance or about the 
characteristics of human exposure to it; and 

(d) the range of uncertainty associated with the estimates. 

4. Quantitative estimates of risk associated with exposure to a substance — 
particularly when expressed in terms of lives likely to be lost or cases of cancer 
likely to occur — have a power to captivate public and press attention. Such 
estimates should be accompanied by statements stressing their imprecision and 
uncertainty. When the available health effects data are seriously deficient or 
little is known about human exposures to a substance, the risk of misinterpretation 
may justify an agency decision not to attempt quantitative estimates of risk or 
health benefits. 

VII. Public Participation 

Several values are served by public participation in the regulatory process. 
Perhaps preeminent is the value of fairness to those who may be directly affected 
by government action. The "right to be heard" before government acts adversely 
to important private interests is well established in American administrative law. 


This is reflected in the Administrative Procedure Act and by most regulatory 

A second purpose served by broad public pariticipation in agency deci- 
sionmaking is avoidance of mistaken even though well intentioned judgments 
that rest on incomplete information. In the context of carcinogen regulation this 
value enjoys a high rank. Decisions concerning human exposure to potential 
carcinogens should rest on sound scientific and economic judgments. Estimates 
of risk and cost require information that agencies often lack, as well as analytical 
skills that are found in industry and many public interest organizations. Ac- 
cordingly, agency procedures should facilitate participation by those in the private 
sector with relevant scientific and economic expertise. 

A third important value served by public participation is balance. Agency 
decisions concerning environmental health hazards require government to take 
large chances with both human life and private business. Data are invariably 
inadequate and estimates of future consequences are problematical; these un- 
certainties simply complicate an already difficult task. Determining the appro- 
priate level of human exposure to a substance reasonably found to cause cancer 
is fundamentally a normative exercise in which there are no experts. Because 
decisions are left to regulators with sometimes scant Congressional guidance, 
agency procedures should facilitate broad participation and vigorous debate to 
assure agency understanding of diverse viewpoints. 

Finally, the opportunity to influence agency thinking, coupled with an 
awareness that agency procedures permit broad participation, contributes to the 
acceptability of agency decisions. 

Public health or scientific interests may not have participated in the regu- 
latory process on equal footing with commercial interests. The extent to which 
this disparity, if it exists, is a function of lack of resources and whether it should 
be addressed through public funding of citizen participation are questions that 
transcend carcinogen regulation. The arguments for and against funding should 
be considered in the broader context. 


1. In setting exposure limits for carcinogens, agencies ordinarily should 
follow procedures that assure opportunities for all affected interests, commercial 
and non-commercial alike, to submit information and views before final decisions 
are reached. Procedures may appropriately vary with the requirements of agency 
organic statutes, with the characteristics of the activity or products whose reg- 
ulation is contemplated, and with the need for prompt action to limit exposure. 

2. Agencies should encourage and facilitate the participation of independent 
experts in toxicology, epidemiology, risk and exposure estimation, and other 
relevant technical disciplines. A useful way of elicitng participation in the reg- 
ulatory process by independent scientists is through the use of standing or ad 
hoc advisory panels. 

3. Congress should refrain from imposing procedural requirements, such 
as section 701(e) of the Federal Food, Drug, and Cosmetic Act, which are so 
burdensome that agencies search for regulatory approaches that have the effect 


of precluding effective participation in their decisionmaking processes by affected 

* See also Administrative Conference Recommendation 71-7 (Rulemaking on a Record by 
the Food and Drug Administration), 2 ACUS Recommendations and Reports 42 (1973); and Rec- 
ommendation 72-5 (Procedures for the Adoption of Rules of General Applicability), 2 ACUS 
Recommendations and Reports 66 (1973), 1 C.F.R. §305.72-5. 




(Adopted December 16, 1982) 

This recommendation focuses on the increasing risk to federal executive 
branch officials of civil liability for monetary damages for alleged violations of 
federal constitutional rights. This vulnerability has expanded dramatically in 
recent years, as a result of judicially-discovered rights enunciated in Bivens v. 
Six Unknown Named Agents, 403 U.S. 388 (1971), and subsequent court cases 
involving allegations of official misconduct. Under the present system of offi- 
cials' liability, as developed piecemeal by the courts, an individual federal em- 
ployee (except certain categories of officials, including the President, who have 
been ruled to have absolute immunity) may be held personally liable for acts 
that, though committed while the employee was acting within the scope of office 
or employment, may subsequently be found to violate a constitutional provision. 
Juries may hold officials liable for actual damages where they cannot show that 
their actions were taken in good faith — that is, in the belief that their conduct 
was lawful — and for punitive damages where they are shown to have acted 
maliciously or with reckless disregard of the plaintiff's constitutional rights. At 
present, damages may not be recovered against the United States for violations 
of constitutional rights as such, although claims arising out of the same conduct 
may or may not be stated against the Government under the Federal Tort Claims 
Act, 28 U.S.C. §§ 2671-2680. 

The existing system of civil sanctions for constitutional violations by federal 
officials does not provide adequate assurance of compensation for victims of 
such violations and discourages proper conduct by government officials. In 
addition, the federal government often has interests at stake in constitutional tort 
litigation involving its officials which may not be represented adequately when 
individual officials themselves are the defendants on trial. 

In Carlson v. Green, 446 U.S. 14 (1980), the Supreme Court suggested 
that the courts may properly refuse to entertain monetary damage actions against 
federal officials if Congress has expressly substituted a different remedy or made 
available an alternative to the Bivens remedy. In the Conference's view, such 
an alternative system is likely to improve the effectiveness with which federal 
programs and laws are administered. 

To serve the primary goals of compensation, deterrence, and fairness in 
dealing with constitutional violations assertedly committed by federal officials, 
and to afford a solution to the problems perceived to flow from the current system 



of individual liability, Congress should replace the existing system by accepting 
public liability for wrongs done in the public's name and by strengthening the 
means of dealing with the wrongdoers. When defending against constitutional 
tort claims, the Government should be able to assert any immunity or good faith 
defense available to the officials. 

Since the Conference's mandate extends only to matters affecting the ad- 
ministration of federal agencies' programs, this recommendation addresses only 
actions against executive agency officials. We do not intend to suggest that the 
same considerations do not apply to officials of the legislative and judicial 


1 . Congress should enact legislation providing that the United States shall 
be substituted as the exclusive party defendant in all actions for damages for 
violations of rights secured by the Constitution of the United States committed 
by federal executive branch officers and employees while acting within the scope 
of their office or employment. The legislation should provide adequate proce- 
dures to ensure that, where a damage action for violation of such rights is brought 
against an executive branch officer or employee, such action should be deemed 
to have been brought against the United States upon certification by the Attorney 
General that the defendant officer or employee was acting within the scope of 
his office or employment at the time of the incident out of which the suit arose. 
The Attorney General's failure to make such certification should be judicially 

2. Such legislation should provide that, in actions alleging constitutional 
violations, the United States may assert as a defense any qualified immunity or 
good faith defense available to the executive branch officer or employee whose 
conduct gave rise to the claim, or his reasonable good faith belief in the lawfulness 
of his conduct. The United States should also be free to assert such other defenses 
as may be available, including the absolute immunity of those officers entitled 
to such immunity. 

3. The agency that employed the offending official should be responsible 
for investigation and, where appropriate, for disciplining the official and imple- 
menting any other appropriate corrective measures. The Office of Personnel 
Management should assure, via guidance promulgated through the Federal Per- 
sonnel Manual and other devices, that agencies are authorized to employ existing 
mechanisms to impose sanctions on officers and employees who have violated 
the constitutional rights of any person. Employees should be permitted to assert 
as a defense in any disciplinary proceeding their good faith in taking the action 
in question, as well as such other defenses as may be available. 

4. Congressional legislation should preserve the opportunity for jury trial 
only with respect to claims that arose prior to the effective date of the legislation 
implementing this recommendation. 


(Adopted December 17, 1982) 

A person adversely affected by an agency rule may ordinarily obtain judicial 
review of that rule either by instituting a direct review proceeding against the 
agency in an appropriate court (pre-enforcement review)' or by asserting the 
invalidity of the rule as a defense in a civil or criminal proceeding to apply or 
enforce the rule (enforcement review). Prior to the Supreme Court's decision in 
Abbott Laboratories v. Gardner, 387 U.S. 136 (1967), direct review was gen- 
erally difficult to obtain because of technical defenses such as lack of ripeness 
or lack of standing, and most review of rules took place in the context of 
enforcement proceedings. 

Under Abbott Laboratories and subsequent decisions, direct review of agency 
rules has become increasingly available. Congress in much recent regulatory 
legislation has specifically provided for immediate resort to judicial review at 
the conclusion of the rulemaking proceeding. As a result, direct judicial review 
of rules has come to be regarded as the norm and review in an enforcement 
proceeding is less common. In a number of statutes, in fact, Congress has sought 
to encourage prompt direct review by explicitly precluding or limiting the avail- 
ability of review at the enforcement stage. 

At the same time, and perhaps largely as a result of the increasing importance 
of direct judicial review of rules, courts have intensified their scrutiny of the 
administrative process preceding promulgation of the rule. Whereas in the pre- 
Abbott Laboratories era challenges to rules were most frequently based on as- 
sertions of lack of agency authority or on inapplicability of the rule to the party's 
particular circumstances, today the issues in direct review proceedings increas- 
ingly include whether the agency made the proper procedural choices in the 
rulemaking proceeding and whether the rule finds adequate support in the ad- 
ministrative record. 

The Administrative Procedure Act does not by its terms establish different 
standards of review for direct review proceedings and enforcement proceedings, 

1 . We use the term ' 'direct review' ' to refer to judicial review of a rule of general applicability 
before the rule is applied to a particular person in an adjudicative proceeding. Such review may be 
by the court of appeals pursuant to a special statutory review procedure or by the district court in 
the exercise of its power under the Administrative Procedure Act to review agency action not otherwise 
reviewable. See ACUS Recommendations 74-4 (Preenforcement Judicial Review of Rules of General 
Applicability) and 75-3 (The Choice of Forum for Judicial Review of Administrative Action). 



and few courts have considered the implications for review in enforcement 
proceedings of the increasingly intensive standard developed in direct review 
proceedings. Moreover, in adopting statutory provisions precluding enforcement 
review Congress has not distinguished between these process-related objections 
and other types of objections to rules. 

In Recommendation 76-4, the Conference criticized provisions precluding 
enforcement review in the Clean Air Act and the Federal Water Pollution Control 
Act. In view of the increasing reliance on direct review and the proliferation of 
issues concerning the adequacy of the rulemaking process rather than the agency's 
authority to promulgate a particular rule, however, the Conference now believes 
that limitations on judicial review of rules in enforcement proceedings may 
sometimes be appropriate. The purpose of this recommendation is twofold: to 
identify factors that Congress should consider in deciding whether to preclude 
enforcement judicial review, and to distinguish between types of challenges to 
rules that should ordinarily be covered by any preclusion provisions Congress 
decides to adopt and types of issues that should ordinarily remain available in 
enforcement proceedings even where preclusion provisions have been adopted. 

Sound principles of administrative law favor prompt and dispositive reso- 
lution of disputed issues arising from an administrative rulemaking proceeding. 
Direct review in the court of appeals is more likely to afford such a resolution 
than later enforcement review in one or more district courts.'^ The uncertainty 
caused by the potential for conflicting court decisions and by the possibility that 
a rule may be overturned several years after its promulgation can be extremely 
disruptive of the regulatory scheme. In addition, reopening a rulemaking pro- 
ceeding to correct any defects will become increasingly difficult as the original 
record grows stale over time and the situation of the interested parties changes. 

On the other hand, those affected by a rule should have a full and fair 
opportunity to challenge the rule on all available grounds. These interests must 
be balanced in determining when limitations on enforcement review of rules 
were appropriate. The balance will tip in favor of limitations on enforcement 
review when the impact of foreclosing review on those affected by a rule is the 
least and when the costs of regulatory uncertainty or of declaring a rule invalid 
after several years are the greatest. Thus one factor favoring limitations on 
enforcement review is the likelihood that the groups affected by rules promulgated 
under a particular statute will be well represented in the agency rulemaking 
proceeding because, for example, those groups are well defined and/or well 
organized. Widespread participation in the rulemaking proceeding reduces the 
probability that any significant issue concerning the rule, particularly one per- 
taining to the rulemaking process itself, will be overlooked on direct review. 

The likelihood that a rulemaking proceeding will involve complex proce- 
dures or intensive factual exploration also militates in favor of limits on enforce- 
ment review. The more elaborate and formal the administrative rulemaking 

2. The recommendation is based on the assumption that, where Congress has provided by 
statute for direct review of rules, that review will ordinarily lie in the court of appeals rather than 
in the district court. See ACUS Recommendation 75-3. 


proceeding (such as those required by hybrid rulemaking statutes^), the more 
Ukely it is that the rule will be subject to challenge on the basis of a relatively 
narrow issue involving the procedures used or the record support for some aspect 
of the rule. As time passes, reopening such a complex proceeding after a court 
reversal will be increasingly burdensome. Encouraging the dispositive resolution 
of challenges to a rule on direct review (perhaps while effectiveness of the rule 
has been stayed) is also advisable when the costs of regulatory uncertainty are 
particularly high. Sometimes compliance with a rule entails substantial expense 
that will not be fully recoverable if the rule is later overturned. Similarly, when 
Congress determines that there is a need to achieve important regulatory goals 
promptly nationwide, or when it is particularly important that rules apply in- 
dustry-wide in order to avoid unfair competitive advantage to noncomplying 
businesses, repeated litigation over the validity of the rules in various district 
courts should be avoided if possible. 

Even when Congress decides that limits on enforcement review are war- 
ranted, it should not foreclose all challenges to rules at the enforcement stage. 
Some grounds for review can be precluded with little unfairness to parties who 
may be unaware of the original rulemaking proceeding or are otherwise unable 
to seek direct review, while others raise fundamental questions about the sub- 
stance of the rule or its application in circumstances that may have been un- 
foreseen at the time of promulgation. Challenges based on asserted errors in the 
administrative process are those most suitable for preclusion. When objections 
on procedural grounds are raised early, errors may be remedied promptly and 
the rulemaking process recommenced with a minimum of disruption to the in- 
terests of those affected by the rule. And objections based on asserted inadequacy 
of the administrative record may lose their relevance as that record itself becomes 
dated. These objections, moreover, do not ordinarily turn on the situation of a 
particular individual or entity or on a particular interpretation of the rule and 
can be raised as well by one party as by another. 

On the other hand, considerations of fairness and judicial economy may 
argue for retaining a right to raise in enforcement proceedings those objections 
based on asserted lack of statutory authority or the inapplicability or unreason- 
ableness of the rule as applied to the facts of the case. Moreover, there may be 
constitutional inhibitions against precluding or restricting at any time challenges 
based on the asserted unconstitutionality of a rule either on its face or as applied. 

The Conference recognizes that the line between issues of process and those 
of statutory authority may not always be a bright one. For example, the question 
of whether there is statutory authority to apply a rule to a particular situation 
may tend to converge with the issue of the adequacy of record support for the 
rule. However, if the distinctions suggested below are made, statutory provisions 
will afford the courts adequate guidance in most situations. When ambiguity 
nonetheless results, the well established presumption of reviewability will con- 
tinue to apply. 

3. See, e.g., ACUS Recommendations 79-1 (Hybrid Rulemaking Procedures of the Federal 
Trade Commission) and 80- 1 (Trade Regulation Rulemaking Under the Magnuson-Moss Warranty- 
Federal Trade Commission Improvement Act). 



1 . In drafting a statute that provides for adequate pre-enforcement judicial 
review of rules. Congress should consider whether to limit the availability of 
review at the enforcement stage. In deciding whether to limit the availability of 
enforcement review in a particular statute, Congress should consider the follow- 
ing factors as favoring such a limitation: 

a) The likelihood that the rulemaking proceeding will attract widespread 

b) The likelihood that the proceeding will involve complex procedures or 
intensive exploration of factual issues; 

c) The likelihood that those affected by the rule will incur substantial and 
immediate costs in order to comply with it; 

d) The need for prompt compliance with the rule on a national or industry- 
wide basis. 

2. When Congress decides to limit the availability of judicial review of 
rules at the enforcement stage, it should ordinarily preclude review only of issues 
relating to procedures employed in the rulemaking or the adequacy of factual 
support for the rule in the administrative record. Judicial review of issues relating 
to the constitutional basis for the rule or the application of the rule to a particular 
respondent or defendant should be permitted when these issues are raised in 
subsequent suits or as defenses to subsequent enforcement actions (subject to 
the principles of collateral estoppel and stare decisis). Judicial review of issues 
relating to the statutory authority for the rule should be precluded at the enforce- 
ment stage only where Congress has concluded that there is a compelling need 
to achieve prompt compliance with the rule on a national or industry-wide basis. 

3. When Congress limits the availability of judicial review of rules at the 
enforcement stage was described in paragraph 2, it should provide that, in an 
exceptional case when foreclosure of issues will work a severe hardship or 
otherwise produce a manifestly unjust outcome, a court may either dismiss or 
stay the proceedings and refer the rule to the affected agency for its reconsi- 

4. Paragraph D of Recommendation No. 76-4, Judicial Review Under the 
Clean Air Act and Federal Water Pollution Control Act, is hereby superseded 
to the extent that it is inconsistent with this recommendation. 




82-5 THROUGH 82-7 



Richard A . Merrill"^ 


This report is a revised and combined version of two chapters of a longer 
report on federal regulation of carcinogenic substances. Readers of this version 
should be aware of my original conception of the full report, which may help 
explain what they read. 

The project reflected in this report was originally commissioned by the 
Office of the Chairman of the U.S. Administrative Conference with only a general 
understanding — on the part of the Chairman's Office or myself — of what form 
it might ultimately take. By the time the project was assigned to a committee 
of the Conference, my conception of the final product had crystallized. I visu- 
alized a report, necessarily very lengthy, consisting of six chapters. The first 
chapter would introduce the subject of carcinogen regulation; the next four would 
deal with four agency programs (discussed in this version); the final chapter 
would attempt a synthesis and evaluation that might or might not support rec- 
ommendations by a Plenary Session of the Conference. 

Work on the report proceeded more slowly than anticipated, for reasons 
that need not be detailed here. By early 1981 , 1 had completed chapter 2, dealing 
with FDA regulation of environmental contaminants of food, and chapter 3, 
dealing with regulation of chronic hazards by the Consumer Product Safety 
Commission. In addition, I had circulated drafts to the committee of chapters 1 
and 4, the latter dealing with OSHA's health standards program. The two com- 
pleted agency chapters have been published in the Virginia Law Review. They 
are FDA Regulation of Environmental Contaminants of Food, 66 Va. L. Rev. 
1357 (1980) and CPSC Regulation of Cancer Risks in Consumer Products: 1972- 
1981, 67 Va. L. Rev. 1261 (1981). Chapter 1 developed into Part I of the present 

At about this time the Conference concluded that recommendations should 
be developed based upon the research already completed and reported (even 
though sometimes not in final form). Accordingly, I was asked to turn to chapter 
6 of my original outline, which focuses on the procedures and institutional 
arrangements through which the four agencies make decisions about potential 
cancer hazards. This chapter appears as Part II of this report. Chapter 4 remains 
in original draft, which is available from the Administrative Conference. Most 
of my research for chapter 5, dealing with EPA's regulation of pesticides, has 
been conTpleted but only a preliminary draft has been submitted to the Confer- 

*= Dean, University of Virginia School of Law. 



I would like to express my appreciation to several students who have given 
generously of their time, and to friends who have read prior versions of one or 
both parts. The students who have contributed importantly include Ursula Bartels, 
Thomas Bartman, Carolyn Flury, Chevis Home, and John Troxel, all of the 
University of Virginia Law School. Margaret Mellon, a former student and 
lawyer with training in biochemistry, was my primary collaborator in preparing 
the treatment of scientific data supporting regulation which appears in Part I. 
Dr. Richard Bates, Joseph Rodricks, Ph.D., Dr. Marvin Schneiderman, Peter 
Barton Hutt, Richard Cooper, and Lester Lave read portions of the two parts 
and provided helpful suggestions. None of them is responsible for any statement 
made or judgment drawn. 

I mvite readers to provide comments and corrections, as well as suggestions 
to amplify the text or supplement current resources. 

Part L Cancer, Causes, and Risk Assessment 

A . Introduction 

Debilitating and often painful, cancer currently is responsible for one-fifth 
of all deaths among Americans' and probably engenders greater fear than any 
other fatal disease.- A major part of the nation's budget for health sciences is 
devoted to investigating cancer and a separate research establishment, the Na- 
tional Cancer Institute, is concerned specifically with finding causes of and 
treatments for cancer.^ In addition, both regulatory agencies and private orga- 
nizations have devoted substantial resources to curtailing human exposure to 
chemicals suspected of posing a risk of human cancer. Four federal agencies 
have been principally involved in regulation of cancer risks — the Environmental 
Protection Agency, the Food and Drug Administration, the Occupational Safety 
and Health Administration, and the Consumer Product Safety Commission."^ 
Each has numerous other responsibilities but rising public apprehension about 
cancer and disappointment with efforts to develop effective treatments led these 
four agencies during the last decade to make regulation of carcinogens a central 
and visible part of their programs.** 

The societal resources devoted to controlling human exposure to toxic sub- 
stances is perhaps the best evidence that regulation of carcinogens has become 
a major activity of government. It is a challenging and difficult activity, both 
for the responsible agencies and for the industrial firms that are the primary 
targets of regulation. A major source of difficulty is the highly imperfect infor- 
mation base for decisionmaking. Three types of uncertainties are persistent. First, 

1. Office of Technology Assessment, Assessment of Technologies for Determining the Car- 
cinogenic Risk from the Environment 31. U.S. Congress (1980) [hereafter OTA Report]. 

2. See S. Sontag, Illness as a Metaphor (1979). 

3. See generally. E. Redman. The Dance of Legislation 29 (1974). 

4. Other agencies have been involved, including the Department of Transportation and the 
Department of Agriculture, but the four mentioned have been principally responsible for regulating 
general public exposure to hazardous chemicals. 

5. See Interagency Agreement: Regulation of Toxic and Hazardous Substances, 42 Fed. Reg. 
54856 (1977). 


the typical candidate for regulation is a substance whose potential role in human 
cancer is inferred from experiments in laboratory animals.^ Theories of carci- 
nogenesis and human experience both suggest that a chemical which induces 
cancer in animals poses a cancer risk for exposed humans, but confirmatory 
evidence for specific substances is often lacking and generally unobtainable. 
Accordingly, regulators, like scientists, must extrapolate from animal data to 
determine human risk. 

A second source of uncertainty stems from disagreement among reputable 
scientists over the meaning of experimental results. The pathological diagnosis 
of tissues of animals used in testing requires both careful observation and ex- 
perienced personal judgment. Pathologists may differ in their evaluation of spe- 
cific tissues or, at the margin, in their diagnostic criteria. Test design may 
occasion disagreement, and often there is some doubt about the care with which 
an experiment was carried out. Study deficiencies or major differences in the 
interpretation of experimental findings can place the issue of a substance's ca- 
pacity to cause cancer in serious doubt. 

A third uncertainty arises from the usual disparity between doses that have 
been associated with cancer induction and the exposure levels that are actually 
confronted by humans or that are achievable without considerable sacrifice. An 
agency may have experimental evidence suggesting that Chemical A induces 
cancer at high concentrations in animals, but it is usually contemplating human 
exposures at a level orders of magnitude lower. Even when this gap does not 
exist, an agency must decide how far exposure must be reduced to protect human 
health adequately. Federal regulators have accepted the premise that no level of 
exposure to a carcinogen can be considered "safe" for all members of an exposed 
population.^ Thus, their ultimate task is to decide what degree of risk can be 
accepted, given the costs of controlling it. 

This question lies at the heart of any decision to control human exposure 
to a potential carcinogen. Existing statutes provide divergent instructions to the 
agencies responsible for answering it. A few statutes, epitomized by the famous 
Delaney Clause,^ can be said to embody Congress' categorical balancing of risks 
and costs. Other statutes, like the Federal Insecticide, Fungicide, and Rodenticide 
Act, and other recent environmental control laws, leave administrators to balance 
risks and costs, with varying degrees of guidance.^ This study explores and 
attempts to evaluate how agencies assigned this large responsibility have exer- 
cised it. Particular attention is devoted to the procedures and institutional ar- 
rangements that they have used in making decisions about cancer risks and the 
costs of reducing them. 

6. For our purposes, a "carcinogen" is a substance that has been found to induce an excess 
of tumors in experimental animals or in humans. 

7. See Scientific Bases for Identification of Potential Carcinogens and Estimation of Risks, 
Report of the Interagency Regulatory Liaison Group, 63 J. Nat. Cancer Inst. 243, 264-65 (1979), 
also reprinted at 44 Fed. Reg. 39857 (1979) [hereafter IRLG Scientific Principles]. 

8. See Food, Drug and Cosmetic Act § 409(c)(3)(A) (codified at 21 U.S.C. 348 (1976)); see 
also Merrill and Hutt, Food and Drug Law, 11-1^ (1980) [hereafter Merrill and Huttj. 

9. See OTA Report at 180-82. 


Part I surveys the landscape of federal regulation of carcinogens. It sum- 
marizes the scientific principles that guide agencies in their identification and 
assessment of chemical carcinogens, and describes the types of evidence on 
which they rely. It explains why current understanding of carcinogenesis forces 
government regulators to balance health risks and regulatory costs in designing 
control strategies for carcinogens, and it reviews the empirical methods used to 
inform this process. The objective is to provide readers with background suf- 
ficient to evaluate the performance of the four agencies, which is assessed in 
Part II and in related studies. 

B. Nature, Incidence, and Causes of Cancer 

It is not the purpose of this section to provide a scientific treatise on the 
etiology or prevalence of human cancer. A rudimentary introduction to the 
underlying mechanisms of the disease will, however, aid understanding of the 
problems confronting regulatory agencies. A summary of recent trends in re- 
ported cancer incidence will enable the reader to evaluate the public health 
significance of government attempts to regulate toxic chemicals. 

Two introductory caveats are to be noted. First, the following descriptions 
of the development and occurrence of cancer are necessarily oversimplified and 
very possibly obscure areas of scientific ignorance or debate. Second, the dis- 
cussion of cancer prevalence and demography, later in this section, should not 
be understood as minimizing the seriousness of the disease or the importance of 
efforts to control it. Rather, the objective is to place in perspective the public 
health implications of regulatory initiatives which are almost always controver- 

1 . Nature of the Disease. 

The term "cancer" describes several similar but distinct diseases, all of 
which are characterized by aberrant and uncontrolled growth of abnormal cells. 
There are three main types of cancers: carcinomas, which invade the body's 
internal and external surfaces and glands and account for over 90 percent of 
cancer deaths; sarcomas, which attack the body's connective tissues and bones; 
and leukemias, which are cancers of the blood cells. '^ All three types develop 
in similar fashion. Cancer cells are descendants of normal cells but their response 
to the body's mechanisms for control of cellular growth has been altered, usually 
by exposure to an exogenous agent.'' Some tumors are benign, i.e., are self- 
contained, and these are usually easily removable by surgery; others metastasize, 
i.e., spread to other parts of the body, and thus are biologically malignant.'^ 
Benign tumors can, but do not invariably, become malignant. Some biologically 
benign tumors, however, may eventually destroy so much adjacent normal tissue 
that they can be characterized as clinically malignant. It is this invasion and 
destruction of normal body tissue in central organs where tumors occur that can 
ultimately cause death of the host. 

10. Cairns, The Cancer Problem, 233 Scientific American 64 (1975) [hereafter Cairns]. 

11. Id. 

12. Id. at 73. 


Cancers probably do not develop in a single step, but through a multi-stage 
process that has two main phases: initiation and promotion. ^^ In the initiation 
stage, an agent induces an irreversible change, probably a mutation, in some 
cells. This event alone, however, may not necessarily lead to the growth of 
malignant tumors. The multiplication of mutant cells into a tumor is the second 
stage of the process, and is often triggered by one or more promoting agents 
which by themselves have little or no capability of causing cancer. ^'^ Initiating 
agents, or initiators, are usually capable of precipitating both phases if exposures 
are large or repeated. Promoters, on the other hand, only act on cells previously 
changed by exposure to an initiator. The promotion stage may not follow im- 
mediately after initiation; there is often a considerable lag time between exposure 
to a carcinogen and development of cancer. 

2. Trends in Cancer Incidence. 

In 1900, approximately 4 percent of all U.S. residents died of cancer. ^^ By 
the mid- 1 930' s, cancer accounted for an estimated 18 percent of all deaths, and 
by 1980 this figure had risen to 20 percent. ^^ The proportion of the population 
dying of cancer has thus risen sharply since 1900; whether the nation is expe- 
riencing an increase in cancer incidence, however, is debated. Reports of rising 
cancer rates triggered public anxiety in the early 1970's,'^ but more careful 
analysis revealed that most of the observed increase was attributable to the rising 
percentage of the population surviving to older ages and to the rise in lung cancer 
resulting from widespread cigarette smoking. ^^ The view persists that age- 
corrected cancer rates have generally remained stable since the turn of the century, 
but apprehensions about a cancer "epidemic" have periodically been revived 
by well-publicized reports. In 1979, for example, a National Cancer Institute 
scientist suggested that increases in age-corrected cancer incidence could not be 
accounted for by cigarette smoking. ^^ The implication was that other causes 
were at work. 

Reports of a rise in cancer have generated intense controversy and have not 
yet gained acceptance. At the base of the disagreements is the fact that reliable 
data for measuring cancer incidence do not exist. In recent years the National 

1 3 . Tumor Promoters: Carcinogenesis Gets More Complicated, 20 1 Science 5 1 5 , 5 1 5 ( 1 978) 
[hereafter Tumor Promoters] . 

14. Id. 

15. See Pollack and Horm, Trends in Cancer Incidence and Mortality in the United States, 
1969-76, 64 J. Nat'l Cancer Inst. 1091 (1980) [hereafter Pollack and Horm]. 

16. Id. 

17. Higginson, Perspectives and Future Developments in Research on Environmental Car- 
cinogenesis, 1:7, 202 Carcinogenesis: Identification and Mechanisms of Action, ed. C. Griffin, and 
C. R. Shaw (1979) [hereafter Perspectives on Environmental Carcinogenesis]. 

18. Id. 

19. Toxic Substances Strategy Committee, Toxic Chemical and Public Protector, a Report to 
the President (Gov't Printing Office, May 1980) [hereafter Toxic Substances Report]. The report 
cites the analysis by M. A. Schneiderman of the cancer incidence data of Pollack and Horm, supra 
note 15. Schneiderman estimated that less than half of the increase in cancer incidence reported by 
Pollack and Horm was due to smoking and suggested that chemical carcinogenesis might account 
for the remainder of the increase. 


Cancer Institute has launched two major survey programs to determine cancer 
incidence: the Third National Cancer Surveys (TNCS), covering 1969-1971, 
and the Surveillance, Epidemiology and End Result Program (SEER), which in 
1973 began to collect cancer incidence and survival data on an annual basis. ^° 
Each survey program covered 10 percent of the U.S. population. The 1979 report 
mentioned above relied on data from these surveys in concluding that cancer 
incidence is rising. -^^ Further analysis of the data was purported to confirm this 
conclusion even when cigarette smoking was controlled for.^^ Critics of the 
report, however, have pointed out that the successive samplings covered differ- 
ent, albeit similar-sized, populations. The observed differences in cancer inci- 
dence could, accordingly, simply reflect differences between the populations 
studied rather than any increase in cancer. -^^ Furthermore, critics have questioned 
the method used by the NCI scientists to account for the large effect of cigarette 
smoking. -^"^ 

In a study that is perhaps the most comprehensive recent effort to assess 
cancer incidence in the United States, British epidemiologists Doll and Peto 
questioned claims that cancer was increasing. The types of cancers being ob- 
served and the trends recorded, they concluded, "do not . . . suggest that the 
United States is beginning to experience an epidemic of cancer due to new 
factors. "-^^ Doll's and Peto's assessment has been endorsed by many health 
leaders in this country. Yet other experts, sometimes relying on different health 
statistics, continue to see evidence of rising cancer incidence that is possibly 
attributable to industrial activity. 

Accurate description of the historical incidence of cancer could facilitate 
an evaluation of the long-term effects of exposure to industrial chemicals. ^^ For 
example, cancer rates before and after large-scale industrialization could be 
compared to estimate the proportion of cancer that might be attributed to exposure 
to industrial chemicals. ^^ Since it is primarily these substances that are subject 

20. See Toxic Substances Report at 146, 149. 

21. 5^^ Pollack and Horm. 

22. See Toxic Substances Report at 160-165. 

23. Smith, Government Says Cancer Rate is Increasing, 209 Science 998, 1000 (1980). In 
a few years, SEER's components will have been standardized for a sufficient time to provide 
dependable cancer trends. In the interim, it seems best to consider the NCI figures to be suggestive, 
rather than determinative, of cancer incidence rates. 

24. See Toxic Substances Report at 161-164; Smith, op. cit. supra note 23, at 1002. 

25. Doll and Peto, The Causes of Cancer: Quantitative Estimates of Avoidable Risk of Cancer 
in the United States Today, 65 J. Nat'l Cancer Inst. 1191, 1258 (1981) [hereafter Doll and Peto]. 

26. Industrialization, however, has also been accompanied by changes in lifestyle that may 
affect human susceptibility to carcinogens and thus the incidence of cancer. This possibility com- 
plicates attempts to attribute changing incidence to industrial chemicals or to see the difference as 
the upper bound estimate of their effects. See Comments of Marvin Schneiderman to the author, 
March 1981 (copy on file at the University of Virginia School of Law) [hereafter Schneiderman 

27. Higginson concludes on the basis of epidemiological studies which have tried but failed 
to detect relationships between cancer incidence and industrialization that "only a small part of the 
cancer burden can be directly related to industrialism in a general sense or to diffuse exposures in 
the general environment. ..." Speech by John Higginson, M.D., Director of the International 
Agency for Research on Cancer, to Joint Board Council Committee on Environmental Improvement 
(1981) [hereafter Higginson Speech]. 


to regulation, such an estimate would establish an upper limit of the potential 
impact of regulation on cancer incidence. -^^ Because the occurrence of cancer in 
any period reflects exposures to carcinogens that commenced years before, ^^ the 
stable rates of most cancers recorded through the 1970's does not demonstrate 
conclusively that increased production of chemicals after World War II has not 
affected cancer incidence. ^° The time elapsed is now approaching the latency 
period for many cancers, however, and data gathered during the current decade 
should help resolve this issue. 

There are two important trends about which there is no serious dispute. 
One is a marked increase over the years in lung cancer, which is attributable 
largely to cigarette smoking;^' the other is a marked decrease in stomach cancer, 
for reasons that are still a matter of conjecture. ^^ Rates for cancers at other sites 
have altered more gradually. For example, NCI estimates that the incidence of 
cervical cancer is decreasing 6 percent annually, while the incidence of melanoma 
skin cancer is rising at a like rate.^^ The fluctuating, patchwork distribution of 
cancer obstructs description of overall cancer incidence, and recent improved 
detection procedures may also make interpretation of general incidence data 
treacherous. The picture is also potentially distorted by the general aging of the 
population,^'* and the decline in incidence of infectious diseases, which once 
competed with cancer as causes of death. ^^ Still another complication is the 
generational association, or "cohort effect," of some cancers. ^^ For example, 
because persons bom early in this century are more likely to have smoked 
cigarettes than persons bom either before or since, the reported rate of lung 
cancer can be expected to increase as more of them reach advanced age. 

28. Even if a relatively certain number could be obtained for the percentage of cancer attrib- 
utable to industrial causes, government cannot feasibly eliminate all exposure to industrial chemicals. 
At most, government agencies can reduce some of the risk due to only a few of the causes of cancer. 

29. See Toxic Substances Report at 1 19-20. 

30. Davis and Magee, Cancer and Industrial Chemical Production 206 Science 1356 (1980). 
The authors caution against relying too heavily on the aggregate production statistics for all organic 
chemicals in pegging the time at which human exposure to harmful chemicals began to increase, 
and consequently, the time to decide whether exposure to industrial chemicals has contributed to 
the cancer burden. First, production levels are only indirectly related to levels of human exposure. 
Second, not all chemicals pose a health hazard and the assumption that the proportion of chemicals 
with high carcinogenesis risk and wide human exposure was the same in 1940 as later may not be 
well founded. According to the authors, aggregate production of synthetic organic chemicals has 
doubled every 7 to 8 years since the early 1940's. Using a 20 to 30 year lag period for the appearance 
of cancer, these data on aggregate production suggest an increase in cancer incidence would have 
been expected in the late 1960's or 1970's. But the authors also compiled individual production 
histories for seven high risk chemicals presently receiving regulatory attention, e.g. acrylonitrile and 
perchloroethylene, and concluded that the first widespread and high level human exposure to these 
substances did not occur until the 1960's. Cancer resulting from these exposures might not be 
expected until as late as the 1990's. 

31. See Toxic Substances Report at 119-20; Cairns, note 10 supra, at 73 (graphs entitled 
Cigarette Smoking and Lung Cancer); OTA Report at 69-72. 

32. Cairns, note 10 supra, at 72. See also Higginson Speech, Slide 17 (stomach cancer 
decreased by 49%). 

33. See Pollack and Horm at 1098-1 100. 

34. OTA Report at 32. 

35. Id. 

36. Id. at 31. 


Thus, the picture of cancer in the United States today cannot be captured 
in a single statistical table. The reported findings that cancer incidence is in- 
creasing annually at a rate of 1.3/1.5 percent for men and 2.0/1.8 percent for 
women, even if they were not disputed, mask counter-trends that site-specific 
analysis reveals. ^^ Furthermore, cancer incidence is rising faster than cancer 
mortality, ^^ indicating that we are having greater success in prolonging lives 
among those who do get cancer even while the disease accounts for a growing 
percentage of all deaths. 

The controversy over whether cancer rates are rising notwithstanding, cancer 
unquestionably is a serious public health concern. Yet it is important to recognize 
that tum-of-the century cancer rates among those who lived long enough to 
exhibit the disease were roughly equivalent to the current rates. ^^ This observation 
suggests that factors in the pre-industrial environment caused cancer with almost 
the same frequency as we experience today, and it should lead us to pause before 
accepting the notion that cancer is primarily a result of industrial activities whose 
by-products can be controlled through regulation. 

3. The Role of Environmental Carcinogens. 

The importance of regulation in reducing the incidence of cancer depends, 
obviously, on the extent to which cancers are caused by substances that can be 
controlled through regulation. If a large proportion of current cancers were caused 
by chemicals added to food, used in consumer products, and encountered in the 
industrial environment, efforts to limit exposure to these substances might prom- 
ise major health gains; if such exposures contribute little to cancer incidence, 
regulation would be less important. Not surprisingly, therefore, the role of en- 
vironmental agents in cancer induction has also been the focus of intense debate. 
There is, as well, confusion about the meaning of the term "environmental." 

A major obstacle to evaluating the role of environmental exposures has 
been our still primitive understanding of the biological origins of cancer. While 
medical science has had some success in developing treatments, its efforts to 
explain the causes of cancer have lagged. These efforts have been complicated 
by the wide variety of cancers, which appear to have different causes and mech- 
anisms. For example, cigarette smoking is the primary cause of lung cancer, 
but does not appear to have any relation to skin cancers, most of which are 
caused by exposure to the sun's natural radiation. A further complication in 
establishing causal relationships stems from the multi-stage, multi-cause nature 
of cancer. "^^ Some substances, such as cigarette smoke and asbestos, are capable 

37. See Pollack and Horm at 1095-97. 

38. Id. at 1095-1100. 

39. See Toxic Substances Report at 147; Perspectives on Environmental Carcinogenesis at 

40. See Tumor Promoters at 515; OTA Report at 66-68. 


of causing cancer independently, but when a person is exposed to them in 
combination, the risk of cancer increases much more than additively."^' In the 
face of this diversity, a simple cause and effect theory of cancer development 
probably cannot be sustained/^ 

While the prevailing belief is that most cancers are induced by "environ- 
mental causes,'"*^ this generalization is easily misconstrued. Some have inter- 
preted the phrase as a shorthand accusation of industrially-related chemical pollution 
as the chief culprit. However, the principal early advocate of the environmental 
hypothesis, John Higginson, has criticized this as an oversimplification.'^'* For 
him, environment is "what surrounds people and impinges upon them.'"^^ Thus, 
in addition to industrial projects and discharges, it includes dietary habits, life- 
styles, preferences such as smoking and sunbathing, place of residence, and 
other circumstances within individual control. "^^ 

The multi-cause nature of cancer and the synergistic properties of many 
carcinogens make determining the percentage of cancer attributable to industrial 
chemicals difficult. The NCI survey purports to show that increases in cancer 
incidence cannot be attributed solely to smoking. "^^ Industry spokesmen, inter- 
preting this finding as an allegation that more cancer is attributable to industrial 
discharges and occupational exposures, have argued persuasively against this 
inference. "^^ Other estimates of the contribution of industrial chemicals to human 
cancer range from as low as 1 percent to over 20 percent. "^^ The latter figure is 
extrapolated from studies of workers exposed to a single substance, asbestos; it 
assumes that, given the number of new chemicals being introduced, other car- 

42. Current theories of carcinogenesis have virtually abandoned the hypothesis that most 
cancers are caused by viruses, though viruses may well play some role in cancer development. See 
OTA Report at 105-06. 

43. Maugh, Cancer and Environment: Higginson Speaks Out, 205 Science 1363 (1979) 
[hereafter Higginson Interview]. Higginson 's "best estimates" attribute 24 percent of the cancers 
in women to known environmental factors like tobacco and sunlight and 63 percent to presumed 
environmental factors of lifestyle. The equivalent percentages for men are 53 and 30 percent. 

44. Id. 

45. Id. at 1363. 

46. Id. 

47. See note 19 supra. 

48. See Morgan, Comment on Papers by Schneiderman and Pollack and Horm, cited in 
"Analysis of Cancer Incidence and Mortality," American Industrial Health Council (1980). But see 
sources cited note 23 supra. 

49. In a public review draft a group of scientists at NCI, NIOSH, and NIEHS estimated that 
20 to 39 percent of cancers are associated with (but not necessarily attributable to) occupational 
exposure. K. Bridbord, P. Decaufle, J. F. Fraumeni, Jr., D. C. Hall, R. N. Hoover, D. P. Rail, 
U. Saffiotti, M. A. Schneiderman, and A. C. Upton, "Estimates of the Fraction of Cancer in the 
United States Related to Occupational Factors" (1978) [hereafter NCI Estimates Paper]. The lower 
estimates for cancers due to occupational exposure were reported by Higginson and Meier, The Role 
of Epidemiology in Elucidating the Importance of Environmental Factors in Human Cancer, I 
Cancer Detection and Prevention 79 (1976), and Environmental Carcinogenesis: Misconception and 
Limitations to Cancer Control, 63 J. Nat'l Cancer Inst. 1291 (1979). 

41. Hammond and Selikoff, Relation of Cigarette Smoking to Risk of Death of Asbestos- 
Associated Disease among Insulation Workers in the United States 312-317, in Biological Effects 
of Asbestos, eds. Bogovski, Gilson, Timbrell, Wagner (1973). 


cinogens of comparable potency will emerge in the future.**" Both this assumption 
and the interpretation of the data on asbestos have been strongly attacked. ^^ 
Higginson and Muir, using a site-by-site approach, and controlling for relevant 
factors such as smoking and "background" incidence, have projected that chem- 
ical/occupational factors may be responsible for as much as 10 percent of lung 
cancers, 30 percent of bladder cancers and leukemias, and all of the rare me- 
sotheliomas.^^ Doll and Peto, in what is probably the most elaborate analysis to 
date, estimate that at most 5 percent of cancers are attributable to manufactured 

The numerous estimates suggest that while industrial chemicals are not the 
causes of most cancers today, their role in the development of certain types is 
probably significant. Furthermore, because of the slow development of cancer, 
there is always the possibility that chemicals now considered safe will later be 
shown to be malign. These considerations support continued activity by the 
agencies responsible for regulating hazardous substances. It is also apparent, 
however, that many so-called "environmental" causes of cancer are controllable 
by individuals, e.g., smoking, alcohol consumption, and dietary patterns. ^"^ In 
addition, some cancers may be attributable to physical factors, such as time 
spent at higher altitudes where ultra-violet radiation is more intense. ^^ Any attack 
on cancer incidence must therefore be broad-based and stress education about 
individual opportunities to reduce risk. And in evaluating the performance of 
regulatory agencies we must recognize that even comprehensive control of ex- 
posure to industrial chemicals probably would not affect most cancers. 

Furthermore, even greater success in controlling causes of cancer might not 
bring about dramatic improvements in aggregate public health. The Conference 
on Death and Dying recently published an analysis, based on published actuarial 
data, of the ultimate goal of regulating carcinogens, viz., the elimination of 

50. The Toxic Substances Report discussed the methodology used in the estimates paper to 
obtain the high values (20-38%) for the occupational contribution to the cancer burden and points 
out that attributing all lung cancers observed in asbestos workers to asbestosis, thus ignoring the 
contribution of smoking, resulted in an overestimate of the number of lung cancers associated with 
asbestos. The report also mentioned several other methodological features which lead to an over- 
estimate of the occupational contribution and still others which suggested that the first number was 
an underestimate. Since the magnitude of these correction factors cannot be determined, the usefulness 
of the NCI Estimates Paper is severely undermined. 

51. See Industry Council Challenges HEW on Cancer in the Workplace, 202 Science 602 

52. Higginson and Muir, Environmental Carcinogenesis: Misconception and Limitations to 
Cancer Control, 63 J. Nat. Cancer Institute 1291 (1979). 

53. Doll and Peto at 1238-45, 1256. 

54. Perspectives on Environmental Carcinogenesis at 194-195. Although diet undoubtedly 
affects cancer incidence, it is difficult to design dietary measures for control of the disease. Diet 
affects human health in numerous and varied ways, and there is some evidence that dietary changes 
that may reduce cancer risk may also increase the risk of heart disease. Elimination of saccharin 
has been claimed to be such a measure. Even the evidence on optimum diet for cancer avoidance 
is rather ephemeral. Schneiderman Comments, supra note 26, at 12. 

55. OTA Report at 102-03, 137-38. 


cancer as a cause of death. ^^ The Conference first compared the main causes of 
death today and those reported at the turn of the century.''^ Numerically, the 
most important causes of death among adults in 1900 were infectious diseases: 
influenza and pneumonia headed the list, tuberculosis was a close second, and 
gastroenteritis ranked third. Cancer ranked well behind, in eighth place. Now, 
by contrast, the three leading causes of death among adults are degenerative 
diseases: heart disease, cancer, and stroke. The changes in the list testify to the 
success of individual lifestyles and public health measures in eliminating infec- 
tious disease, which now rank fifth among causes of death. ^^ Degenerative 
diseases typically materialize late in life. The shift to heart disease and cancer 
as leading causes of death accordingly parallels an increase in average life 
expectancy. ''^ 

Although these data reveal success in postponing the average occurrence 
of death to older ages, there is little evidence that the biological life span of 
humans has been increased. Recognizing that all humans will eventually die, 
the Conference study analyzed the effects of eliminating cancer as a cause of 
death. The study's central conclusion was surprising: "Actuaries tell us that if 
cancer were to be eliminated as a cause of death, average life expectancy would 
be increased by 2.5 years for those under 35. . . .At age 65 the expectation of 
life, if cancer were eliminated, would increase by only 1 .4 years. "^^ Obviously, 
individuals "saved" from cancer would succumb to death from some other cause. 
Making the assumption that those spared would succumb to other causes of death 
in proportion to their current numerical importance, the authors of the study 
estimated that 193,000 more people would die of heart disease, 45,000 of strokes, 
and 20,000 of accidents.^' 

These computations dramatize the significant difference in public health 
terms between eliminating a disease in 1980 and eliminating a disease in 1900. 
Eradication of infectious diseases early in the century both eliminated a prominent 
cause of death and dramatically extended human life expectancy. ^^ Today, be- 
cause the major fatal diseases strike so late in life, the elimination of one disease 
has the effect of substituting another cause of eventual death without comparable 
effects on average life span. Of course, causes of death differ sharply in their 
accompanying consequences, such as individual suffering and the costs of treat- 
ment. Moreover, small gains in life expectancy spread over a large population 
may seem very significant. Nonetheless, it is instructive to realize that the 
aggregate effect of eliminating cancer would be to "expose" most current cancer 

56. If Nobody Died of Cancer . . . ? An Analysis of Demographic Data. Conference on Death 
and Dying: An Examintion of Legislature and Policy Issues, ed. J. Dinstusie (1976). 

57. Id. at 12. 

58. Id. at 13. 

59. Id.atW. 

60. Id. at 13. 

61. Id. at 14. 

62. Id. at 11-12. 


victims to death from other diseases within a relatively short time.^^ And since 
the probable causes of most of those deaths also often require extensive medical 
treatment, total health care costs might not shrink dramatically.^ 

C. The Regulatory Dilemma 
1. Tenets of Carcinogen Regulation. 

Although the choices confronting agencies responsible for regulating car- 
cinogens may not be unique, the common necessity to balance incommensurate 
values imposes special difficulties. Designing controls for any carcinogen inev- 
itably involves a tradeoff, implied or express, between health benefits and other 
values. The necessity to choose stems from the tenet that no level of a carcinogen 
can be considered safe. It has generally been assumed that any exposure to a 
carcinogen, no matter how small, entails a small but greater-than-zero risk of 

This assumption contrasts with the premises that have historically governed 
regulation of toxic materials. Classical toxicology holds not only that the sub- 
stance's toxicity is a function of the dose, but also that there is some dose below 
which an exposed organism will not experience any adverse effects. ^^ In other 
words, some level of the substance can be considered literally "safe." To 
determine such a level for a conventional toxicant, the regulator requires the 
substance to be tested in experimental animals to determine the dose level at 
which no adverse effects are observed. To this "no observed effect level" he 
then will apply a so-called safety factor, i.e., a discount, of 10, 100, or occa- 
sionally even 1000, to allow for the genetic variability of the population.^ 

While this approach has been questioned, ^^ it is in principle consistent with 
prevailing theories about the occurrence of toxic effects. Conventional toxicants 
are assumed to have a threshold because the kinds of effects they produce occur 
only when enough cells are functionally impaired to produce an adverse effect 
on the organism as a whole. For example, most individuals can sustain damage 
to a few liver cells with no ill effects, because undamaged cells will take over 
their functions. It is only when many of the cells in a specific organ are damaged 
that the body is unable to compensate and permanent harm results. ^^ Furthermore, 

63. The time may not be nearly as short as the cited figure suggests, however. If only 20 
percent of the population die from cancer, elimination of the disease should yield 12.5 additional 
years of life for those who now die of cancer, i.e., 2.5 years times 5. Schneiderman Comments at 

64. Conference on Death and Dying Report at 15. 

65. Zbinden, The No-Effect Level, an Old Bone of Contention in Toxicology, 43 Archives of 
Toxicology 3 (1979). 

66. See 21 C.F.R. § 170.22 (1978); R. Merrill, Regulating Carcinogens in Food: A Legis- 
lator's Guide to the Food Safety Provisions of the Federal Food, Drug, and Cosmetic Act, 11 Mich. 
L. Rev. 171, 207 (1979) [hereafter Merrill, Carcinogens in Food]. See also Schneiderman and 
Brown, Estimating Cancer Risks to a Population, 22 Environmental Health Perspectives 115, 117 

67. See Zbinden note 65, supra. 

68. Kimm, Kuzmach, and Cotruvo, The Questionable Value of Cost-Benefit Analysis: The 
Case of Organic Chemicals in Drinking Water 4-5, (1979). 


the traditional "safety factor" approach often provides the basis for an agency 
to defend an exposure limit that still permits use of a toxic substance. If the 
level estimated to be "safe" cannot be achieved without major sacrifice, of 
course, the agency may have to moderate its insistence on safety. The important 
point, however, is that the scientific premises underlying regulation on conven- 
tional toxicants do not inexorably force a compromise between enhancement of 
human health and other societal objectives. 

Precisely such a trade-off is dictated by the prevailing tenets governing the 
regulation of substances that are believed to pose a risk of cancer. For such a 
substance, whether identified by animal experiments or through analysis of hu- 
man exposure, ^^ regulators have assumed that no finite level of exposure can be 
considered safe for all individuals. This "no threshold" assumption is not strictly 
a statement of scientifically demonstrated fact; rather it represents a regulatory 
policy based on both hypothesis and prudence. ^° 

The "no threshold" policy has been supported, first, by uncertainty about 
how cancer occurs, coupled with the hypothesis that even a single molecule of 
a carcinogen can initiate cancer in a vulnerable individual. Understanding of the 
process of chemical carcinogenesis remains incomplete, but cancer is believed 
to begin in the DNA of a single cell which becomes transformed into a tumor 
cell that can multiply uncontrollably and eventually kill the host. By whatever 
means chemical carcinogens participate in that process, only one molecule or a 
few molecules are likely to be involved in the initial reaction with cellular 
components. ^^ The likelihood that any molecule will reach and interact with the 
critical DNA site is small, however. For this reason, the risk that cancer will 
actually develop at low exposure levels is itself low, but it increases with in- 
creasing doses. 

The "no threshold" assumption has also been supported by the inability 
of laboratory scientists or epidemiologists to demonstrate a no-effect level for 
carcinogens.^^ The small size of standard animal experiments make it impossible, 
even when no positive response is obtained at a given dose, to rule out the 
possibility that a larger sample would have revealed such a response. Enlargement 
of a study population can reduce, but never eliminate, this possibility of a false 
negative. ^^ The numbers of humans at risk in most epidemiological studies are 
typically much larger, but the problem remains. Furthermore, interpretation of 
negative findings is complicated by lack of good information about dose levels, 
duration of exposure, and individual health histories.^'* 

69. See text accompanying notes 198-284, infra. 

70. For a statement of the threshold concept accepted by four major agencies regulating 
carcinogens, see IRLG Scientific Principles at 264-65. 

71 . For that single molecule to reach the vulnerable point in the cell, another complex process 
of metabolic transformation and transport must occur. See OTA Report at 66. 

72. See text accompanying notes 222-48, infra. 

73. See text accompanying notes 234-35, infra. 

74. See OTA Report at 40-42 (discussion of the errors which may inhere in relying on death 
certificates as a source of raw data for the calculation of cancer incidence). 


It is for these reasons that regulatory agencies have accepted as an article 
of faith that any level of exposure to a substance capable of causing cancer 
carries some risk, i.e., the possibility that one or more members of the exposed 
population may get cancer. An important, less universally accepted corollary is 
that the risk presented by an initiator carcinogen is roughly proportional to dose. 
This corollary is supported by limited human evidence and consistent with the 
behavior of most agents identified as carcinogens in animal experiments. Several 
biostatistical models have been developed to define the relationship, i.e., to 
estimate the shape of the dose-response curve for individual carcinogens. While 
these models have different rationales and sometimes project widely varying 
dose-response relations, most assume that the risk of cancer disappears only 
when the dose of a carcinogen reaches zero.^^ 

For many years a third operating assumption has enjoyed acceptance among 
regulatory agencies: All carcinogens were alike; no distinctions could be dem- 
onstrated among agents associated with an increase in tumors in man or in 
animals that justified different regulatory outcomes. ^^ The reluctance of regu- 
lators to draw such distinctions stemmed initially from a lack of theoretical bases 
for distinguishing among carcinogens based on the way in which they interact 
with mammalian cells. Scientists have more recently come to recognize a variety 
of mechanisms by which agents contribute to the development of cancer. ^^ 
Generally, however, neither experimental evidence nor clinical studies have yet 
provided convincing evidence for characterizing the mechanism of specific car- 
cinogens in regulatory proceedings.^^ 

Regulatory agencies accept a fourth premise that largely determines the type 
of evidence on which they rely. They believe that the experience of animals 
exposed to a substance provides good evidence for predicting its effects in 
humans. This correlation underpins modern toxicology and pharmacology, and 
has been accepted by regulators for decades in estimating both acute and chronic 
toxicity in humans. The long latency of most cancers, coupled with requirements 
for premarket evaluation of safety, have made animal experiments the principal 
source of data for regulatory decisions concerning potential carcinogens. 

The central tenets of regulation can thus be summarized: (1) Any substance 
that induces cancer in animals presents a probable risk of cancer for man. (2) No 
dose of such a substance can be characterized as completely "safe." (3) The 

75. Schneiderman and Brown, Estimating Cancer Risks to a Population, 22 Environmental 
Health Perspectives 115, 118-19(1978). 

76. See IRLG Scientific Principles at 246. 

77. See Tumor Promoters, note 13, supra. 

78. There have nonetheless been suggestions that science should develop the bases for reg- 
ulatory distinctions between promoters and initiators, accompanied by recommendations that reg- 
ulators should act on them. See, e.g., Kolbye, Impact of Short-Term Screening Tests on Regulatory 
Action, in Applied Methods in Oncology No. 4, Elsevier Biomedical Scientific Publishing Company, 
Amsterdam { 1 980); Potter, Initiation and Promotion in Cancer Formation: The Importance of Studies 
on Intercellular Communication, 53 Yale Journal of Biology and Medicine 367 (1980). While 
available information about chemical carcinogens is usually insufficient to show that they act through 
a mechanism for which there is a threshold, in a few cases physical mechanisms have been dem- 
onstrated. Schneiderman Comments at 20. 


risk of cancer can, however, be reduced by limiting human exposure. This does 
not mean that agencies routinely regulate all animal carcinogens as cancer risks 
to humans or automatically seek to prevent any exposure whatever. Agencies 
have sometimes declined to treat a substance that causes cancer in animals as 
posing a risk of cancer for man because of differences in route of exposure or 
in metabolism. ^^ Furthermore, an agency may conclude that the risk associated 
with very low levels of exposure to a recognized carcinogen is so slight as to 
be of no concern, and thus "safe" in some colloquial sense. Nonetheless, the 
summary correctly states the prevailing premises of federal regulation, which 
define the dilemma confronting regulatory agencies: If any exposure to a car- 
cinogen carries some risk of cancer and reducing exposure will diminish the 
risk, how shall an agency determine the level of exposure to be permitted? 

2. Typology of regulatory approaches. 

At least two regulatory issues must be resolved to justify government action 
to regulate human exposure to a chemical. First, it must be determined whether 
the chemical is capable of harming persons who may be exposed. And, second, 
it must be determined whether humans are likely to be exposed to the chemical. 
In the absence of affirmative answers to both, government intervention to control 
exposure would be difficult to justify. A very few statutes require only these 
two findings as prerequisite for regulation. An example is the Food Additives 
Amendment to the Federal Food, Drug, and Cosmetic Act^^ which requires FDA 
to disapprove, i.e., prevent human exposure to, any food additive shown to be 
capable of causing cancer. Indeed, under the Amendment FDA rarely need 
trouble with the issue of exposure, which ordinarily is taken as conceded by the 
filing of a petition for marketing approval. ^' 

Most laws under which carcinogens are regulated, however, demand that 
the agency consider other criteria. These other criteria, though multifaceted, can 
be summarized under two headings: the magnitude of the risk posed by a chemical 
and the effects of regulating it. One can also visualize a decisional formula that 
required an agency to consider only the former of these criteria, in addition to 
toxicity and exposure. Thus, government might undertake to control exposure 
to chemicals that pose a risk greater than some specified level, but not to weigh 
any costs of control. Such a standard would force the decisionmaker to find 
methods for quantifying risk, but it is not as demanding — of thought or infor- 
mation — as a standard that requires consideration of the effects of controlling 
exposure. It thus leaves unexplored issues that are relevant to rational regulatory 

79. The nutrient selenium has been associated with an increase in cancers of the liver in 
experimental animals, but FDA has explicitly authorized its addition to both human and animal 
food. 38 Fed. Reg. 10458 (1973); 39 Fed. Reg. 1355 (1974). The premise for the agency's decision 
was that the substance was not a direct-acting carcinogen, but promoted the development of tumors 
by damaging liver tissue. See Bureau of Food "Double Standard" on Selenium Alleged, Food 
Chemical News, July 6, 1981, at 28. 

80. See note 8 supra, and accompanying text. 

81. See generally Merrill, Carcinogens in Food at 204-09. 


It is difficult to visualize a standard that required an agency to consider the 
issues of toxicity, exposure, and costs of control, but not the magnitude of the 
risk. Yet most statutes fail to address this subject, and some early decisions 
appear consistent with the belief that the magnitude of the risk either did not 
matter or could not be determined. In recent years, however, agencies have 
sought ways of quantifying the risk posed by carcinogens.^^ If an agency must 
consider other values in deciding whether or how to regulate, quantification 
becomes critical. Without some quantitative measure, however rough, of the 
health risks presented by various levels of exposure to a carcinogen, an agency 
has no way of determining whether control measures are too costly or too lenient. 

The criteria for regulatory decisionmaking embodied in current legislation 
fall into three broad categories. A very few statutes mandate a "no risk" policy, 
which means that the administering agency must attempt to prevent human 
exposure to any carcinogen. Other laws, either expressly or through adminis- 
trative construction, focus exclusively on the issue of risk but essentially direct 
the agency to regulate only significant risks. A third, much broader class of laws 
direct the administering agency to allow other criteria to moderate its efforts to 
protect human health. These criteria may be limited to the technological capability 
of an industry to reduce exposure, or they may include a wider range of con- 
sequences of regulation. While the laws within this last category vary widely, 
they all in some fashion permit regulators to determine how much safety should 
be required. 

This question is central to the decisions of EPA, FDA, OSHA, and the 
CPSC that are the subject of this study. The laws they administer do not provide 
the same answer. ^-^ Indeed, it has been suggested that current legislation reflects 
no coherent Congressional policy toward cancer risks, but rather conveys quite 
disparate and sometimes even contradictory directives to regulators.^'* The reality 
is less disconcerting. 

To be sure, the nearly thirty statutes^^ under which Federal agencies regulate 
carcinogens seem to express different degrees of concern about risks to human 
health and about the weight to be given economic costs in determining control 
levels for toxic agents. This diversity has several explanations. The laws were 
enacted at different times. More recent statutes reflect increased Congressional 
sensitivity to economic costs and describe in more detail the factors an agency 
shall consider in setting control levels. ^^ The laws originated with, and remain 
under the influence of, different political constituencies. One would expect that 
legislation drafted by the House Agriculture Committee would reflect greater 
concern about economic costs of restrictions on pesticide use than laws designed 
to protect worker health sponsored by the Senate Committee on Labor and Public 
Welfare. Perhaps most important, none of the laws was enacted for the sole 

82. See text accompanying notes 290-341 infra. 

83. OTA Report at 176-181. 

84. "A Low Growth Microcosm," Wall Street Journal, October 10, 1976. 

85. OTA Report at 178-179. 

86. See. e.g.. Consumer Product Safety Act, 15 U.S.C. §§ 2651-2082 (1976 and Supp. Ill 
1979); Toxic Substances Control Act, 15 U.S.C. §§ 2601-2631 (Supp. Ill 1979). 


puq)ose of controlling cancer risks; rather each is addressed to a particular 
medium or commercial activity that may present several kinds of health hazards. 
None of the laws examined in this study expressly mandates formal bal- 
ancing of health benefits and costs; Congress has not directed any agency to 
perform the kind of rigorous cost-benefit analysis that many economists es- 
pouse. ^^ More commonly, it has prescribed regulation of "unreasonable" risks, 
sometimes specifying the consequences that it wants the agency to consider and 
occasionally intimating that the agency should not impose controls that are too 
costly. The very looseness of Congress' language in many laws suggests that, 
at least in the majority of cases, there are not insurmountable legal impediments 
to adoption of consistent policies by the four agencies. 

3. Analysis of current approaches. 

With this introduction, it is useful to consider the prototype approaches in 
more detail. It should be noted that the following discussion does not purport 
to supply definitive interpretations of the statutes discussed. Rather, the purpose 
is to describe and evaluate prototype approaches to the regulation of chemical 

(a) No risk. This approach is epitomized by the famous Delaney clause 
of the Federal Food, Drug, and Cosmetic Act. The clause was enacted in 1958 
as part of the Food Additives Amendment, which requires that any food additive 
be found "safe" before FDA may approve its use in food.^^ The Delaney clause 
stipulates that this finding cannot be made for a food additive that has been 
shown to induce cancer in man or in experimental animals. Congress has thus 
precluded FDA from permitting any level of human exposure to a carcinogenic 
food additive. The statute does not require, nor has it been interpreted as per- 
mitting, FDA to make any further inquiry to determine, e.g., whether some 
finite level of a carcinogen may be considered "safe enough" in light of the 
benefits of its use.^^ A similar approach ostensibly is embodied in the provision 
of the Clean Air Act governing toxic air pollutants. ^^ 

The "no risk" approach is not typical of federal environmental or health 
laws. Indeed, the FD&C Act embodies disparate food safety standards. One 
provision, under which FDA has regulated environmental contaminants of food, 
directs the agency to consider offsetting factors such as the price and availability 
of food.^' Furthermore, the draconian impact of the Delaney Clause has occa- 
sionally led FDA to decline to categorize a carcinogen as a food additive, to 

87. Legislation explicitly mandating cost-benefit analysis is rare. See American Textile Mfgs. 
Inst. V. Donovan, 452 U.S. 490, 509-12 (1981). 

88. 21 U.S.C. § 348(c)(3)(A). 

89. See Merrill, Carcinogens in Food at 179-184. 

90. Clean Air Act § 1 12, 42 U.S.C. § 7401 et seq. (1976); see National Emissions Standard 
for Identifying, Assessing, and Regulating Airborne Substances Posing a Risk of Cancer, 44 Fed. 
Reg. 58641 (1979). 

91. 21 U.S.C. § 346. See Merrill and Schewel, FDA Regulation of Environmental Contam- 
inants of Food, 66 Va. L. Rev. 1357 (1980) [hereafter Merrill and Schewel, Environmental Con- 


avoid having to ban a food substance that provides substantial benefits. ^^ 

FDA's efforts to escape the Delaney clause illustrate the principal difficulty 
with a "no risk" approach to regulating carcinogens: its propensity to produce 
unacceptable results.'^'* The clause has been characterized as a categorical cost- 
benefit judgment by Congress, i.e., a judgment that no food additive is likely 
to offer benefits sufficient to outweigh any risk of cancer. "^"^ Recent FDA efforts 
to regulate such substances as saccharin and nitrite, however, have confirmed 
the practical limitations of such categorical judgments. '^^ 

(b) Negligible risk. The risk posed by a carcinogen depends on the dose; 
smaller effective doses present lower risks. It may be possible to reduce human 
exposure to some carcinogens to so low a level that the associated risk, though 
greater than zero, is small enough to ignore. Where Congress found such low 
levels achievable for a class of products, it could direct an agency to allow their 
use if any associated risk were extremely small, without requiring the agency 
to take into account control costs. 

No current health law mandates such a "negligible" or "insignificant" risk 
approach. FDA has, however, adopted it administratively for three classes of 
products. '^^ FDA's approach represents an administrative effort to interpret delphic 
legislative instructions in light of rapidly advancing technology. It is best illus- 
trated by the agency's proposed criteria for regulating carcinogenic animal drugs 
under a 1962 exception to the Delaney clause. ^^ To understand FDA's approach, 
one must be familiar with this exception. 

After 1958, the Food Additives Amendment was interpreted as prohibiting 
the approval, for use in animals produced for human food, of any drug that had 
been shown to cause cancer; this was a "food additive" use barred by the 
Delaney clause. In 1962 Congress amended the Delaney clause to permit FDA 
to approve the use of a carcinogenic animal drug if the agency were convinced 
that "no residue" of the substance would be "found" in edible tissues of the 
treated animals. ^^ The amendment specified that FDA was to prescribe the 
analytic method for verifying the absence of residues. On its face, this amendment 
could be viewed simply as a clarification of the pre- 1962 law: No carcinogenic 
"food additive" could be approved, but if a substance did not remain in human 
food, the policy of Delaney did not preclude its approval. Progressive improve- 
ments in analytical chemistry have since confirmed that no animal drug is entirely 

92. See R. Kingham, Statutory and Administrative Theories by Which FDA Avoids Applying 
the Delaney Clause (Nov. 10, 1977); Merrill, Carcinogens in Food. 

93. Examples arguably include FDA's attempted ban of saccharin, its successful banning of 
cyclamate, and the threatened elimination of sodium nitrite as a food preservative. 

94. Merrill; Regulation of Toxic Chemicals, Book Review, 58 Tex. L. Rev. 463, 474 (1980). 

95. See, e.g., Crandall and Lave, eds.. The Scientific Basis of Health and Safety Regulation 
117-170 (1981); Lave, The Strategy of Social Regulation 11-13 (1981). 

96. These are animal drugs, hair dyes, and contaminants of color additives. In addition, the 
agency has reportedly contemplated the same approach for indirect food additives. See Policy for 
Regulating Chemicals in Food and Color Additives, 47 Fed. Reg. 14464 (1982); Merrill and Hutt 
at 497. 

97. See Merrill, Carcinogens in Food at 227-235. 

98. See 21 U.S.C. § 360b(d)(l)(H); Merrill, Carcinogens in Food at 227 et seq. 


eliminated; a few molecules will always remain in the animals even if they 
escape detection by existing analytical methods. FDA recognized, as the statutory 
language suggested, that the issue was the detection, not the occurrence, of 
residues. It became obvious that sooner or later some analytical method would 
be capable of identifying residues of any drug administered to food producing 
animals. ^^ 

To escape this inexorable conflict, FDA has proposed criteria for approving 
methods to search for drug residues. '^^ The agency proposed that any analytical 
method approved for controlling a carcinogenic drug be capable of measuring 
residues presenting more than an ''insignificant" risk of cancer. And it went 
further to specify, as a quantitative measure of "insignificance," a 1 in 1,000,000 
lifetime risk of cancer. If a drug sponsor provided a detection method capable 
of measuring residues posing any greater risk,'^' FDA in effect said it would 
ignore residues presenting a lower risk.'^^ 

FDA has followed or considered essentially the same approach for other 
classes of products. It has approved a carcinogenic color additive used in hair 
dyes on the premise that the cancer risk associated with any amounts absorbed 
through the scalp, as estimated by quantitative risk assessment, will not exceed 
1 in 1 ,000,000. '^^ And it is considering adopting the same standard for controlling 
both indirect food additives, e.g., packaging materials, and trace contaminants 
that are often unavoidable by-products of the processes used to make direct food 
and color additives. '^"^ In the latter two cases human exposure to the carcinogenic 
material is undesired; rather it is the result of the material's use for other pur- 
poses. '^^ 

FDA's "insignificant risk" approach differs from the Delaney Clause be- 
cause it contemplates that humans may knowingly be exposed to some quantity 
of a carcinogen. The approach also differs from the various trade-off approaches 
discussed below in that it does not require the agency to weigh factors other 
than the substance's carcinogenic potency and the ability of analytical chemists 
to measure residues, i.e., to control exposure. '^^ The approach does not ignore 

99. See Hess & Clark v. FDA, 495 F.2d 975 (D.C. Cir. 1974), analyzed in Merrill and Hutt 
at 485 et seq. 

100. The agency first published its proposal in 1973, 38 Fed. Reg. 19226, and final regulations 
in 1977, 42 Fed. Reg. 10412. These regulations were set aside on procedural grounds and a new 
proposal was published in 1979. 

101. Chemical Compounds in Food-Producing Animals: Criteria and Procedures for Evalu- 
ating Assays for Carcinogenic Residues, 44 Fed. Reg. 17070 (1979). 

102. This is not literally true, but functionally accurate. The 1979 proposal made clear FDA 
would not adopt more sensitive assay methods as they were developed if existing methods could 
detect any residues of concern. 

103. 45 Fed. Reg. 72112 (1980). 

104. See Food Chemical News, April 7, 1980; see also Merrill and Hutt at 95. 

105. In this respect, the substances are like pesticide residues. Lead acetate, on the other 
hand, is the essential ingredient in hair dyes marketed for persons who are concerned about greying 

106. See Diethylstilbestrol: Withdrawal of Approval of New Animal Drug Applications; 
Commissioner's Decision, 44 Fed. Reg. 54852 (1979); Merrill, Carcinogens in Food at 234. 


the accepted premises of carcinogen regulation, but neither does it force the 
agency to evaluate consequences, such as direct control costs or forgone eco- 
nomic benefits, that might logically prompt a decision not to restrict exposure 
to a carcinogen. 

FDA's "insignificant risk" approach does, however, have limitations. First, 
its implementation depends on the availability of data concerning the potency 
of a carcinogen. This is not an insuperable problem for FDA, because the agency 
can insist that a drug's sponsor conduct the testing necessary to generate the 
necessary data.'^^ But agencies responsible for regulating chemicals whose oc- 
currence is not intentional lack comparable leverage. Second, FDA's approach 
requires the selection of one or more quantitative methods for estimating the 
cancer risk associated with low doses of a carcinogen. '^^ Most important, as a 
practical matter, a "negligible risk" approach can work only where human 
exposures to a carcinogen are naturally low. When human exposure is the by- 
product of cattle production or the use of food packaging materials or derma- 
tological products, it may be possible to achieve very low effective doses without 
major sacrifice. But in many contexts humans are exposed to much higher levels, 
e.g., in the workplace, which cannot be easily reduced. 

(c) Trade -ojf approaches. This heading embraces a variety of legal for- 
mulas that have important operational differences but one common feature: each 
requires the regulatory agency to investigate and weigh factors in addition to the 
health risk posed by substances targeted for regulation. These factors offset the 
desire to eliminate exposure to substances that increase the risk of human cancer. 
Three current laws illustrate the variety of trade-off approaches, as well as their 
difficulties. ^^^ 

(i) OSHA's ''feasibility" standard. In setting workplace health standards 
for toxic materials, OSHA is directed to select the standard "which most ade- 
quately assures, to the extent feasible . . . that no employee will suffer material 
impairment of health or functional capacity.""^ On the premises described 
above, assurance that "no employee" would get cancer would require elimi- 
nating exposure to any carcinogen. Historically, however, OSHA has considered 
two offsetting factors. It has interpreted the "feasibility" standard as requiring 
it to consider the availability of technology for reducing exposure, and the 
financial ability of the responsible industries to pay for the necessary controls. ' ' ' 
At the same time, the agency has declined to take into account the financial 
survival of individual firms, and it has expressly refused to balance the health 
benefits of mandated exposure controls against the total costs of achieving them."^ 

107. See 21 C.F.R. §§ 348. 360b (1976). 

108. It would be possible for an agency to use several different methods to illustrate the range 
of possible risks, or select the one in each case which yielded the highest estimated risk. But 
adherence to some method is a precondition to this approach. 

109. There are other such laws. See OTA Report at 178-179. 

110. 29 U.S.C. § 655 (b)(5) (1976). 

111. See American Textile Mfrs. Ass'n v. Donovan. 452 U.S. 490, 495 (1981); Berger and 
Riskin, Economic and Technological Feasibility in Regulating Toxic Substances Under the Occu- 
pational Safety and Health Act, 7 Ecology L.Q. 285 (1978). 

1 12. See Industrial Union Dept., AFL-CIO v. Hodgson, 499 F.2d 467 (D.C. Cir. 1974). 


Readers familiar with OSHA's health standards program will recall the 
controversy engendered by its interpretation of the Occupational Safety and 
Health Act."^ This controversy twice came to the Supreme Court. On the first 
occasion, the Court ruled that before ordering reductions in worker exposure to 
a toxic chemical, OSHA must show that existing levels posed a "significant" 
health risk.''"^ In the second case challenging OSHA's standard for cotton dust, 
a non-carcinogen, the Court rejected claims that the agency is required to balance 
compliance costs against health benefits."^ The result of these two rulings is to 
establish a two-tiered formula for regulation. OSHA must initially find that the 
risk posed by a toxic material is significant; once past that threshold it need only 
consider the technological feasibility and affordability of new controls. 

(ii) FDA's "avoidability" standard. In addition to regulating substances 
used as ingredients in fabricated foods, FDA controls human exposure to sub- 
stances that contaminate agricultural commodities as the result of industrial 
accident or environmental exposure. Prominent examples include mercury in 
seafood, PCBs in poultry and fish, and aflatoxins on peanuts. The applicable 
statute specifies that the agency "shall" establish tolerances for added poisonous 
substances that are "unavoidable through good manufacturing practices.""^ 
This provision does not direct FDA to eliminate contaminants or require their 
reduction to the level necessary to assure that no consumer will suffer harm. 
Rather, it instructs the agency to set tolerances that take into account both the 
objective of protecting health and the extent to which contaminants can be 
avoided. Since most contaminants cannot be reduced without rendering food 
unmarketable, the decision for FDA is how much contamination to allow. "^ 

To answer this question, FDA evaluates the effects of different potential 
tolerance levels on the price and availability of contaminated food."^ For ex- 
ample, FDA has declined to order the destruction of all peanuts containing 
detectable levels of aflatoxin, a potent carcinogen, because of the economic 
importance of this commodity. ' '^ Thus, in regulating carcinogenic contaminants, 
FDA has moderated its search for safety by weighing other values. The agency 
has insisted publicly that it will not take into account effects on the economic 
well-being of food producers '^° and it has declined to weigh such "secondary 
economic effects" of regulation as the impact of its PCB tolerance on interests 
dependent on sport fishing in the Great Lakes area. '^' The levels of contamination 
that the agency permits, however, often are high enough to reflect consideration 
of these values. 

1 13. See generally. J. Morran. OSHA After Ten Years (paper prepared for AEI Conference 
on Health, Safety, and Environmental Regulation, November 18-19, 1981). 

1 14. Industrial Union Dep't, AFL-CIO v. American Petroleum Inst., 488 U.S. 607 (1980). 

1 15. American Textile Mfrs. Ass'n v. Donovan. 452 U.S. 490 (1981). 

116. 21 U.S.C. § 346. See also Merrill and Schewel, Environmental Contaminants. 

117. /i/. at 1379-82. 

118. Id. at 1380. 

119. Id. at 1403 et seq. 

120. Id. at 1992-98; see also Poisonous or Deleterious Substances in Food, 39 Fed. Reg. 
42743 (1974). 

121. Polychlorinated Biphenyls; Residues in Human Food. 44 Fed. Reg. 38330 (1979). 


(iii) EPA' s "unreasonable adverse effects" standard. The Federal Insec- 
ticide, Fungicide, and Rodenticide Act (FIFRA)'^^ requires that any pesticide 
must be registered by EPA before it can be marketed. EPA is directed to refuse 
or withdraw registration if it finds that use of a pesticide is likely to result in 
"unreasonable adverse effects on health or the environment."'^'' The agency 
interprets this language as requiring that it weigh all of the effects of a pesticide — 
its contribution to food production as well as its possible adverse effects on 
applicators, consumers of food, and the natural environment — in determining 
whether, or on what terms, to permit registration.'^"^ EPA has exhibited special 
concern about pesticides that pose a cancer risk,'^^ but it has not taken the 
position that human exposure must be eliminated regardless of offsetting ben- 
efits. '^^ 

FIFRA comes as near as any statute to requiring a comprehensive evaluation 
of risks and benefits of a health-threatening activity, but it is not unique. '^^ 
EPA's experience in administering FIFRA illustrates both the virtues and dis- 
advantages of this approach. Its decisions explore a wide variety of effects of 
pesticide use and restriction, and usually avoid all-or-nothing outcomes. At the 
same time, EPA's procedures for evaluating the risks and benefits of pesticide 
use are more protracted than those of other agencies. For example, before EPA 
reaches a final decision about continued registration of a potential carcinogen, 
it consults with an in-house scientific group, '^^ solicits the advice of the De- 
partment of Agriculture and a statutory advisory committee, '^^ provides at least 
two opportunities for the submission of written evidence,'''^ and often conducts 
a formal evidentiary hearing that may itself require as long as a year.'"" The 
complexity of EPA's process is not attributable solely to the decisional criteria 
it must apply, but the agency's obligation to entertain evidence and argument 
on the benefits as well as the risks of pesticides is surely a contributing cause. 

(d) Impact of current approaches. Each of the foregoing approaches is 
compatible with the prevailing tenets of carcinogen regulation. All four agencies 
accept the premise that no finite level of exposure to a carcinogen can be con- 

122. 7 U.S.C. §§ 135-136(1976). 

123. 7 U.S.C. § 135a (c)(5). 

124. See Regulations for the Enforcement of the Federal Insecticide, Fungicide, and Roden- 
ticide Act, 40 Fed. Reg. 28242 (1975); see also 45 Fed. Reg. 52628 (1980) (proposed revisions). 

125. 40 Fed. Reg. 28242. 28265 et seq. (1975). See. e.g., EDF v. Ruckelshaus. 439 F.2d 
584 (D.C. Cir. 1970); EDF v. EPA, 510 F.2d 1292 (D.C. Cir. 1975); EDF v. EPA. 548 F.2d 998 
(D.C. Cir. 1977). 

1 26. It is uncommon, however, for EPA to sanction continued use on food crops of a pesticide 
that has been shown to be carcinogenic. 

127. See OTA Report at 178; Toxic Substances Control Act, 15 U.S.C. §§ 2601-2629 (Supp. 
Ill 1979). 

128. This is the Cancer Assessment Group. See conversation with Elizabeth Anderson, Di- 
rector, EPA Cancer Assessment Group, Oct. 13, 1981. 

129. See 7 U.S.C. § 136d(a). 

130. See 7 U.S.C. § 135d(d). 

131. C. Home, EPA's August 1980 Proposal; Reforms in the Rules of Practice for Cancellation 
Hearings at 16-18 (Unpublished paper, January 1982). 


sidered safe for all individuals. ^^^ But with the exception of the "no risk" 
standard of the Delaney Clause, each of these approaches allows human health 
to be sacrificed to other objectives. For several cancer risks, such as environ- 
mental contaminants of food and occupational carcinogens, forbidding exposure 
is not a realistic option, even though not literally unachievable. But even when 
it would be relatively easy to eliminate human exposure, e.g., to pesticides and 
animal drug residues, regulators temporize. It should hardly be surprising that 
laws designed to protect health also seek to accommodate conflicting social 
objectives. Experience suggests that the "no risk" approach breaks down when- 
ever the perceived benefits of a product or activity appear substantial. The 
"negligible risk" approach is similarly limited because it demands that exposure 
levels be extremely low. As will be evident, however, the more appealing "trade- 
off" approaches themselves entail significant costs. 

4. Agency Programs Studied for Report. 

The previous section has provided a typology of legal approaches to reg- 
ulating carcinogens. This section briefly describes the regulatory programs se- 
lected for study, all of which fall within the "trade-off" category. The four 
programs are FDA's regulation of environmental contaminants of food; CPSC's 
chronic hazards program; OSHA's regulation of workplace carcinogens; and 
EPA's regulation of pesticides. Each program will be examined in detail in an 
independent chapter. The summaries that follow stress the legal standards under 
which regulatory decisions are to be made and the administrative procedures 
Congress has prescribed for making them. 

(a) FDA Regulation of food contaminants. Under the Federal Food, Drug, 
and Cosmetic Act (FDC Act),'^^ FDA is generally responsible for assuring the 
safety of human food. The statutory provisions that govern this activity are the 
product of a series of Congressional enactments spanning more than six de- 
cades. '^"^ In consequence FDA evaluates the safety of food under several legal 
standards which apply to specific categories of foods (e.g., agricultural com- 
modities) or food constituents (e.g., food additives, pesticide residues). 

Because of the notoriety of the Delaney clause, it is sometimes assumed 
that the FDC Act prohibits any carcinogen in human food. In fact, the clause 
itself applies to relatively few categories of food substances, although FDA has 
applied the "no risk" policy that it embodies more generally. ^^^ As often, 
however, the agency has refrained from interpretations of the Act that would 
bar the sale of food which contains carcinogens. One example is the category 
of agricultural commodities, many of which naturally contain constituents that 
have been shown experimentally to cause cancer. '^^ Another is environmentally 

132. See IRLG Scientific Principles at 259. 

133. 21 U.S.C. § 321 et seq. (1976). 

134. See Merrill, Carcinogens in Food at 184. 

135. Merrill and Hutt at 55. 

136. Id. at 53-56. For a discussion of the number of natural food constituents that have been 
found carcinogenic in animal tests, see Citizen Petition of the Grocery Manufacturers of America 
to the U.S. Food and Drug Administration (August 20, 1981), at 25-27 [hereafter GMA Petition]. 


contaminated food which, as noted above, FDA has regulated under statutory 
language that enables it to weigh values that in its judgment may override the 
objective of minimizing the risk of human cancer. It is this program that 1 have 
selected for special attention. 

Environmental contamination is a significant source of toxicants in food. 
Contaminants occur unavoidably in the harvesting or production of many com- 
modities, such as grains, milk, and fish. Many of these substances are carcin- 
ogenic. Under section 406 of the FDC Act'"*^ FDA is empowered to establish 
tolerances for food contaminants — even carcinogens — if they cannot be avoided 
by good manufacturing practice. In addition to considering public health, the 
agency has taken into account a variety of values, including the cost of imple- 
menting more rigorous controls and the impact of lower tolerances on the avail- 
ability and price of food. It has also given weight to a third criterion: the ability 
to measure the contaminant. The agency cannot enforce a tolerance below the 
level that the best practicable method of chemical analysis can detect. 

The FDC Act prescribes on-the-record rulemaking for establishing toler- 
ances for food contaminants.'^^ Because of the costs of this procedure, however, 
FDA has usually eschewed formal tolerances and instead announced "action 
levels" — contamination levels that will cause the agency to initiate court en- 
forcement under the general adulteration provisions of the Act.'^^ These action 
levels thus operate as informal tolerances, and FDA sets them by considering 
the same criteria as are prescribed by section 406. Action levels are not, however, 
the product of public rulemaking. 

(b) CPSC regulation of chronic hazards. Of the four agencies studied, 
the Commission has played the smallest role in Federal efforts to control exposure 
to toxic chemicals. The agency was established by the consumer Product Safety 
Act of 1972, '"^^ with authority to regulate consumer products that pose an "un- 
reasonable risk of injury or illness." Consumer products include all articles sold 
for general or personal use in the home, school, or recreational settings, apart 
from products regulated by FDA, tobacco, and certain other items.''*' The CPSC 
also administers the older Federal Hazardous Substances Act.'"^^ Together, these 
statutes give the agency two primary tools for controlling human health hazards: 
the power to prescribe mandatory performance and labeling standards for con- 
sumer products; and the power to ban products whose risks cannot be adequately 
controlled through such requirements.'"*^ 

137. 21 U.S.C. § 346 (1976). 

138. 21 U.S.C. § 371(e) (1976). 

139. Specifically, section 401(a)(1), 21 U.S.C. § 341 (a)(1). See Merrill and Schewel, En- 
vironmental Contaminants at 1382. 

140. 15 U.S.C. § 2051 et seq. (1976 and Supp. Ill 1979), as amended. 

141. 15 U.S.C. § 2052 (a)(1) (1976). 

142. 15 U.S.C. § 1261 et seq. (1976). 

1 43 . Under section 1 5 of the CPS Act , the Commission also has the power, after an evidentiary 
hearing, to order the repurchase, recall, or making of refunds for, any consumer product that presents 
a "substantial hazard." 15 U.S.C. § 2065 (1976). Under the shadow of this authority, the agency 
has forced manufacturers to cease distribution of many assertedly hazardous products including ones 
which ostensibly posed a risk of cancer (hairdryers containing asbestos). This authority has been 
used only once to regulate a carcinogen. 


Section 8 of the CPS Act authorizes the Commission to ban a consumer 
product if it finds that "no feasible . . . standard . . . would adequately protect 
the public from [an] unreasonable risk of injury."''*'* The FHSA empowers the 
CPSC to regulate products that are ''toxic," defined by Congress as having the 
capacity to produce personal injury or illness through ingestion, inhalation, or 
absorption.''*^ While the FHSA is primarily a labeling statute, it also authorizes 
the Commission to ban any hazardous substance that presents so serious a hazard 
that "the objective of the protection of the public health and safety can be 
adequately served only by keeping such substance . . . out of the channels of 
interstate commerce. "'"^^ The agency has relied on both statutes to eliminate 
carcinogens from consumer products. 

In assessing the need for a product safety rule or ban for a potentially 
hazardous consumer product, the CPSC is required by statute to weigh the degree 
and nature of the risk, the number of consumer products affected, consumer 
needs, and alternative methods for minimizing the risk.''*^ The legislative history 
of the CPS Act indicates that the CPSC should balance the likelihood that a 
product will cause harm and the gravity of such harm against the effect of 
regulation on the product's utility, cost, and availability to consumers.''*^ Recent 
court cases hold that this formula requires a balancing of economic cost and 
health benefits. '"^^ The FHSA makes clear that before banning a potentially 
hazardous product, the Commission should explore the less stringent alternative 
of labeling, but it does not specify what weight the agency should accord factors 
other than the risk to health in deciding whether a ban is appropriate.'^^ 

While neither statute mentions the risk of cancer, since 1975 the Commission 
has devoted special attention to chronic health hazards. In 1978 the agency 
announced its intention to prevent known carcinogens from being intentionally 
added to consumer products if they could be absorbed, inhaled, or ingested into 
the human system. '^' Under this policy, use of any carcinogenic substance would 
be phased out unless no reasonable substitute were available and elimination of 
the substance would result in "unacceptable" economic and social costs. In the 
latter event CPSC would require reduction of the substance to the lowest attain- 
able level until a substitute is identified. While the CPSC's formal policy was 
set aside on procedural grounds, '^^ the approach it announced continued generally 
to reflect agency philosophy. 

Whether proceeding under the CPS Act or the FHSA, the CPSC has the 
burden of justifying any restriction on the marketing of a product. The law does 

144. 15 U.S.C. § 2057(d) (1976). 

145. 15 U.S.C. § 1261(g) (1976). 

146. 15 U.S.C. § 1261(g)(1) (1976). 

147. 15 U.S.C. § 2058(c) (1976). # 148. See Merrill, CPSC Regulation of Consumer Prod- 
ucts Posing a Risk of Cancer 1972-1981, 67 Va. L. Rev. 1261, 1279^80 (1981) [hereafter Merrill, 
Consumer Products] . 

149. See Aqua Slide' n Dive v. CPSC, 569 F.2d 831 (5th Cir. 1979). 

150. Merrill, Consumer Products at 1290-95. 

151. 43 Fed. Reg. 25658 (1978). See Merrill, Consumer Products at 1297-1300. 

152. See Dow Chemical, U.S.A. v. CPSC, 459 F. Supp. 378 (S.D. La. 1978). 


not obligate manufacturers to notify the Commission of their plans to market 
new products or to obtain prior approval for any design or constituent. The CPS 
Act prescribes the same procedures for adopting a consumer product safety 
standard and for banning a product, i.e. quasi-formal rulemaking. '^^ When seek- 
ing to ban a substance under the FHSA, however, the Commission must follow 
the formal rulemaking procedures set forth in section 701 of the FDC Act.'^"^ 
Because such rulemaking is likely to be protracted, the agency customarily 
proceeds under the CPS Act.'^^ 

(c) OSHA regulation of workplace carcinogens. OSHA's efforts to limit 
worker exposure to toxic chemicals have helped make it one of the most con- 
troversial federal agencies. ^^^ The agency's statute, the Occupational Safety and 
Health Act of 1970 (OSH Act),'^'' subjects employers to two primary obligations: 
(1) a general duty, enforceable administratively, to provide employment that is 
"free from recognized hazards that are . . . likely to cause death or serious 
physical harm . . .";'^^ and (2) the obligation to comply with safety and health 
standards, which constitute the primary means by which OSHA has attempted 
to limit worker exposure to toxic chemicals. The OSH Act specifies that, in 
controlling worker exposure to a toxic material, OSHA shall prescribe the stan- 
dard "which most adequately assures, to the extent feasible, on the basis of the 
best available evidence, that no employee will suffer material impairment of 
health or physical capacity. . . ."'^^ 

The criteria OSHA must follow in setting health standards are a product 
both of statutory language and judicial interpretation. The courts of appeals have 
acknowledged OSHA's broad discretion to prescribe standards even in the ab- 
sence of definitive evidence on health effects.'^ The Supreme Court, however, 
has held that OSHA must show, by substantial evidence, that current exposure 
to a substance poses a "significant" health risk before it can order new con- 
trols. •^' The agency has consistently acknowledged that it must take into account 
whether particular safeguards are technologically and economically "feasible," 
but simultaneously insisted that a "feasible" standard may mandate the devel- 
opment of new technology or even force some firms out of business. This view 
has now been essentially endorsed by the Supreme Court, which agreed that 
OSHA is not obligated to show that the health "benefits" of a standard are 
worth the cost.'^^ 

153. 15 U.S.C. § 2058(d)(2). 

154. 21 U.S.C. § 371(e), incorporated by reference in 15 U.S.C. § 1261(g)(2). 

155. Merill, Consumer Products at 1289-90. 

156. See, e.g. , Nichols and Zeckhauser, Government Comes to the Workplace: An Assessment 
of OSHA. 54 Pub. Int. 39, 55-57 (1979). 

157. 29 U.S.C. § 651 et seq. (1976). 

158. 29 U.S.C. § 654 (2) (1976). 

159. 29 U.S.C. § 655(b)(5) (1976). See American Textile Mfgrs. Institute v. Donovan, 452 
U.S. 490(1981). 

160. See Industrial Union Dept., AFL-CIO v. Hodgson, 499 F.2d (DC. Cir. 1974); see also 
Berger and Riskin, supra, note 1 19. 

161. Industrial Union Dept., AFL-CIO v. American Petroleum Inst., 448 U.S. 607 (1980). 

162. American Textile Mfrs. Ass'n v. Donovan, 452 U.S. 490 (1981). 


In early 1980, OSHA formally promulgated criteria for identifying, clas- 
sifying, and regulating potential carcinogens.'^^ A major objective was to fo- 
restall relitigation of basic scientific principles in individual standard-setting 
proceedings. The agency simultaneously established a system under which it 
would rank individual chemicals as candidates for regulatory action based on 
the strength of the evidence of their carcinogenicity.'^ This novel experiment 
in generic rulemaking is examined in Part II of this report. 

OSHA has the burden of initiating action to control worker exposure to 
carcinogens; no employer need obtain prior approval for any workplace condition 
or practice or, as a legal matter, test new materials before introducing them. 
The procedure for establishing permanent health standards is a variety of quasi- 
formal rulemaking, which provides a public hearing as well as the opportunity 
to submit written comments. Hearings are conducted before an Administrative 
Law Judge, and witnesses are ordinarily subject to questioning by other partic- 
ipants.'^^ By contrast with the CPSC's rulemaking hearing, those at OSHA may 
consume several weeks. 

(d) EPA regulations of pesticides. EPA regulates the marketing and use 
of pesticides and shares with FDA responsibility for regulating pesticide residues 
on food. Pesticide legislation ranks among the oldest Federal laws designed to 
control chemical hazards. '^^ The current law, the Federal Insecticide, Fungicide, 
and Rodenticide Act (FIFRA),'^^ was enacted in 1947 but has experienced major 
amendment in 1964 and on three different occasions in the 1970's.'^^ Under 
FIFRA no pesticide may be marketed unless it has been registered, i.e., licensed, 
by EPA. Section 3 of FIFRA specifies that a pesticide shall be registered if its 
composition and labeling meet the Act's requirements, it will perform its intended 
function, and "when used in accordance with widespread and commonly rec- 
ognized practice it will not generally cause unreasonable adverse effects on the 
environment."'^^ The statute further defines the criterion "unreasonable adverse 
effects on the environment" as "any unreasonable risk to man or the environ- 
ment, taking into account the economic, social and environmental costs and 
benefits of the use of any pesticide. ' ' ' ''^ Thus, in determining whether to register 

163. Identification, Classification, and Regulation of Potential Occupational Carcinogens, 45 
Fed. Reg. 5001 (1980) [hereafter OSHA Carcinogen Policy]. 

164. Id. at 5205-07. This portion of OSHA's so-called "generic cancer policy" has been 
stayed as the agency, under new management, has invited further public comment on the 1980 
regulations. 46 Fed. Reg. 187 (1982). 

165. See R. Merrill, OSHA Regulation of Occupational Carcinogens 27-30, Chapter 4 of 
Federal Regulation of Cancer-Causing Chemicals, a report to the U.S. Administrative Conference 
(Draft, October 1980) [hereafter OSHA Draft]. 

166. The first federal pesticide legislation was enacted in 1910, four years after the 1906 
Food and Drugs Act. See generally, Merrill, EPA Regulation of Carcinogenic Pesticides 2, Chapter 
5 of Federal Regulation of Cancer-Causing Chemicals, a report to the U.S. Administrative Conference 
(Draft, August 1 98 1) [hereafter EPA Draft]. 

167. 7 U.S.C. §§ 135-136 (1976 and Supp. Ill 1979). 

168. EPA Draft at 14. 

169. 7 U.S.C. § 136a(c)(5)(D). 

170. 7 U.S.C. § 136(bb). 


a pesticide, EPA must balance its health and environmental risks against the 
benefits of its use. 

To aid in determining whether a pesticide poses "unreasonable adverse 
effects," EPA has, by regulation, established a series of "rebuttable presump- 
tion[s] against registration." '^' EPA's announcement of a rebuttable presumption 
or RPAR for a pesticide sets in motion an elaborate public process, consisting 
of two opportunities for written submissions, consultation with an advisory com- 
mittee, and sometimes a formal evidentiary hearing to determine registrability. 
EPA has used the RPAR process primarily in conjunction with its review of 
pesticides registered many years ago under less stringent standards than the 
current law imposes. Under EPA's regulations a rebuttable presumption against 
registration arises if a pesticide exceeds criteria for any of three types of risks: 
acute toxicity; lack of emergency treatments for exposed humans; or chronic 
toxicity. '^^ With respect to the latter risk, EPA regulations specify that a pre- 
sumption against registration arises if a pesticide "induces oncogenic effects in 
experimental mammalian species or in man as a result of oral, inhalation or 
dermal exposure. . . ."'^^ 

EPA has also published criteria for assessing whether any agent within its 
jurisdiction poses a cancer risk to humans. '^"^ The criteria recognize four grades 
of evidence of potential harm carcinogenicity ranging from the "best evidence" 
(consisting of epidemiological studies in humans in conjunction with confir- 
matory animal tests), through "substantial evidence" (consisting of animal tests 
that demonstrate tumor induction), to least persuasive "ancillary evidence" 
(consisting of studies that relate chemical structure to carcinogenicity).'^^ Even 
if a pesticide were shown by the "best evidence" to be a human carcinogen, 
however, FIFRA theoretically would permit it to be registered if EPA determined 
that the economic benefits of its use outweighed the health risk.'^^ 

EPA's regulation of pesticides is the only one of the four programs studied 
that incorporates a premarket clearance feature. Before the agency will approve 
a pesticide, the applicant must submit evidence of its efficacy, its environmental 
chemistry, and its degree of hazard to humans, domestic animals, and nontarget 
organisms. '^^ Most of the data underlying EPA's decisions are provided by 
pesticide sponsors, and its regulations state emphatically that the burden of 

171. 40C.F.R. § 162.1 1 (1981). These regulations, first promulgated in 1975. were an attempt 
to orchestrate public debate over the risks and possible offsetting benefits of hazardous pesticides 
outside the context of a formal evidentiary hearing. See 40 Fed. Reg. 28281 (1975). 

172. 40C.F.R. § 162.11(a)(3) (1981). 

173. Id. at § 162.11(a)(3)(ii). 

174. 40 Fed. Reg. 21402 (1976). The criteria appeared in the "Notice" portion of the Federal 
Register, not as regulations. They outline a "weight of the evidence" approach to evaluation of 
suspect compounds, rather than purporting to accord decisive weight to particular test findings. 

175. Id. at 21404. 

176. There are few examples of such decisions, but they do exist. EPA has very rarely allowed 
continued use of a carcinogenic pesticide on human food crops which might bear residues that 
consumers would ingest. On the other hand, EPA has more often continued to permit marketing 
and use of carcinogenic agents where only applicators have any risk of exposure. 

177. 7 U.S.C. § 136a(c). 


proving that a pesticide will not cause "unreasonable adverse effects" rests at 
all times on the registrant. '^^ 

The statutory procedure for pesticide registration — and cancellation — re- 
sembles formal adjudication and includes an opportunity for a formal evidentiary 
hearing. '^^ Hearings are never demanded when EPA denies initial registration, 
but registrants are very likely to insist on a formal hearing when the agency 
attempts to cancel registration of products already in use. EPA's formal deci- 
sionmaking process, accordingly, is the most elaborate of those examined in this 

D. Regulatory Assessment of Risk 

1. Overview. 

The first stage of any decision to regulate a potential carcinogen is an 
assessment of the health risk it poses. Determining risk to health is the central 
element of most decisions by the agencies studied and it would be a major part 
of any cost-benefit assessment. Harm to human health is the most notable adverse 
consequence of allowing exposure to a carcinogen; the principal "benefit" of 
regulating exposure can therefore be expressed in terms of reduced likelihood 
of cancer. 

Assessing the health risk posed by a substance routinely involves at least 
two, and often three, stages of fact-finding and analysis. '^° The risk posed is a 
function of the chemical's ability to cause harm and the type and extent of human 
exposure to the chemical. The manner of exposure can be important because 
only certain types or routes of exposure may be capable of causing harm. The 
level of exposure is important because a substance may be harmful only when 
ingested or inhaled in large quantities or for long periods. Even if a substance 
is capable of causing harm at very low exposure levels, determining actual levels 
of exposure is important in assessing both the size of the risk presented and the 
controls that can be achieved. 

The initial step in risk assessment is to make a qualitative judgment about 
a chemical's carcinogenicity. ^^^ In deciding whether a chemical is capable of 
causing cancer, an agency may draw on four types of evidence: (1) data from 
epidemiological studies in human populations; (2) the results of experiments 
involving laboratory mammals; (3) the results of so-called "short term tests" in 
lower-order organisms, usually bacteria, and (4) comparison of a chemical's 
structure to the structure of oth^r chemicals whose biological activity is well- 

178. 40 C.F.R. § 162.6(b)(2)(B). 

179. 7 U.S.C. § 136d(d). 

180. For a more thorough discussion of the elements of agency decisionmaking, see Part II 
of this report. 

181. The discovery that a chemical has the capacity to cause cancer has usually been the 
starting place for regulatory analysis and theoretically is anterior to the other steps. In some contexts, 
however, an agency may sensibly begin at another point. For example, the CPSC has sometimes 
focused first on the exposure issue. It has attempted to determine the composition of consumer 
products whose use could present serious risks if they contained a hazardous material, and then 
considered whether any of the components are toxic. See Merrill, Consumer Products at 1306-09. 


established. Since only one of these types of data involves exposures of primary 
interest, i.e., human exposures, the appropriateness of extrapolating other types 
of evidence to humans is a persistent issue. 

The second step in risk assessment is to determine whether and how humans 
might be exposed to the chemical. Both facets of this question can be complex. 
For a chemical used as a food additive, human exposure is easily verified; the 
substance is intended to be ingested by humans. For many chemicals, however, 
human exposure is not desired, and in some cases may be unexpected. The 
exposure of milk drinkers to pesticides used on oranges whose rinds become 
forage for dairy cattle is an example. '^^ Confirming that exposure occurs is only 
part of the inquiry, however, for humans may not encounter the chemical in a 
way that permits absorption or by the route shown experimentally to cause 
cancer. ^^^ Furthermore, the quantity of a substance that is encountered and 
absorbed, i.e., the effective dose, will decisively influence the magnitude of 
risk but is often difficult to determine. 

A third step in risk assessment is to determine the magnitude of the cancer 
risk associated with predicted exposures to a carcinogen. As observed earlier, 
this step may not be a prerequisite to regulation. For example, in deciding whether 
to ban a food additive FDA under current law need know only that the additive 
induces cancer in laboratory animals or in humans; it need not measure the level 
of use or determine whether the risk is large or small. '^"^ In many programs, 
however, a qualitative determination of carcinogenicity is not a sufficient basis 
for regulation; ^^^ some agencies may be required to quantify the risk associated 
with exposure to a carcinogen. If an agency is required to consider other offsetting 
values, quantification of the risk assumes added significance. Some attempt to 
gauge the magnitude of the risk posed by a chemical has thus become a common, 
if not routine, step in regulatory risk assessment. '^^ 

The following sections explore in more detail the sources of information 
and the types of analyses that agencies rely on in assessing the health risks of 
potential carcinogens. 

2. Epidemiological Studies. 

a. Background. Among methods for detecting environmental hazards to 
human health, epidemiology ostensibly is the most powerful and yet often the 
least revealing. This dichotomy is sharpest in the case of diseases that become 

182. See NAS. Regulating Pesticides 166-74 (1980). 

183. That human exposure is by a different route does not negate the existence of human 
hazard, but it does pose an additional issue to be resolved in evaluating the degree of hazard. See 
FDA's discussion of the risk posed by lead acetate in hair dyes, 44 Fed. Reg. 12205 (1980). 

1 84. See text accompanying notes 89-95 supra; Food, Drug, and Cosmetic Act § 409(c)(3)(A), 
codified at 21 U.S. C. § 348 (1976); Merrill and Hutt at 77-78. It may do so, of course, for other 
purposes. In proposing its ban of saccharin, for example, FDA presented a fairly lengthy analysis 
of the exposure issue. See 42 Fed. Reg. 1996-2010 (1977). 

185. For example, the Supreme Court's ruling that OSHA must find risk "significant" before 
it can establish a standard is thought by many observers to require quantification. See Industrial 
Union Dep't, AFL-CIO v. American Petroleum Institute, 448 U.S. 607 (1980). 

186. See text accompanying notes 285 et seq. infra. 


manifest only long after primary exposure, of which the clearest example is 

"Epidemiology" embraces a variety of methods for investigating the oc- 
currence and the causes of human disease at the population level. The methods 
used in epidemiological research are essentially descriptive rather than experi- 
mental, i.e., they gather and interpret data from past or continuing human ex- 
posures. ^^^ Epidemiological studies can provide two useful types of information 
concerning cancer. First, they help characterize cancer as a disease, helping 
answer such questions as: Who gets cancer? At what age? Is incidence increasing 
or decreasing? Second, epidemiological methods can reveal the existence, or 
more commonly define the likelihood, of strong associations between environ- 
mental factors and cancer. Studies designed to derive the latter kind of infor- 
mation are more frequently encountered by regulators. 

Epidemiologists gather information describing the diseases experienced by 
a population and information about the behavior or environment of that popu- 
lation. Their operating assumption is that the two types of information can 
illuminate the causes of disease. But because epidemiologists cannot manipulate 
the conditions in which people actually live, they use methods that attempt to 
deduce causes of cancer from available data that reveal associations between 
disease and other factors. '^^ In addition to searching for associations, epide- 
miologists are called on to interpret those that are observed. An observed as- 
sociation may reflect nothing more than a chance relationship between the factor 
and the disease or, at the other extreme, it may signify that the factor itself 
causes the disease. 

In epidemiology, "cause" is defined operationally: An association between 
a factor and a disease that is "strong enough" is deemed to be causal. Whether 
an association is strong enough depends not only on its statistical significance 
but on other, nonquantitative criteria. For example, the appearance of the disease 
must follow the occurrence of, or exposure to, the putative cause. The absence 
of other more convincing explanations for the association is also supportive. 
Most important, the assertion of a causal relationship must be consonant with 
existing knowledge about how the disease occurs. 

A demonstrated association between exposure to a chemical and an increased 
incidence of a type of cancer would be itself be only weak evidence of a causal 
relationship. The inference is, however, strengthened by general acceptance of 
a mechanism by which some chemical exposures can lead to the appearance of 
cancer. The prevailing theory of chemical carcinogenesis hypothesizes a chemical 

187. The epidemiologist often starts from a hunch, perhaps suggested by medical experiments 
or by a preliminary review of existing data, about what environmental factors might influence the 
appearance of cancer. He then analyzes the data, trying to discern associations that are consistent 
with the suspected relationships between factors and diseases. Convincing demonstrations of asso- 
ciation require accurate observations of both disease incidence and other potentially casual factors. 
Although the data available to epidemiologists are often voluminous, they are rarely satisfactory. 
OTA Reports at 136-140. 

188. Id. 


interaction with mammalian cells, in which they are altered from a normal to a 
cancerous state. A strong association between a type of cancer and exposure to 
a specific chemical may therefore be evidence that the chemical is a cause of 
that type of cancer. The probative force of such an association is a function of 
two criteria. One is how well the characteristics of the observed association fulfill 
the biologically-based expectations of causal relationship. The other is the com- 
patibility of the observations with other possible explanations for the associations. 
For reasons described below, available human data ordinarily are not adequate 
to establish conclusively such causal relationships; neither are capable of ruling 
out specific chemicals as causes of human cancer. '^^ 

b. Standard approaches. Epidemiologists follow two basic approaches in 
investigating the connection between environmental insults and cancer: cohort 
studies and case-control studies.'^ Cohort studies, which may be either pro- 
spective or retrospective, compare the health histories of groups of individuals 
who differ in their exposure to the chemical under study. For example, a ret- 
rospective cohort study might compare the health histories of workers in a 
shipbuilding facility that used asbestos insulation to the histories of other workers 
who had no exposure to asbestos.'^' To the extent possible, workers would be 
subdivided according to their level of exposure to asbestos. The incidence of 
lung cancer among workers in each group would be compared to identify as- 
sociations between the level of asbestos exposure and the occurrence of disease. 
A cohort study can convincingly attribute the differences in cancer rate to asbestos 
exposure only if all of the cohorts are well matched with respect to other variables, 
e.g., sex, occupation, age, and exposure to other carcinogens such as cigarette 
smoke, that may influence the incidence of the disease. This ideal is very difficult 
to' achieve. 

The second common format is the case-control study. Such a study begins 
with the identification of individuals who suffer from a disease. These "cases" 
are then compared with matched ''controls," i.e., individuals who do not have 
the disease, to determine whether there are notable differences in exposure to 
suspect agents. As in cohort studies, the objective of the case-control approach 

189. The reliability of a finding of a causal relationship between exposure to a substance and 
cancer is obviously important to environmental regulation. For only if a substance in some way 
contributes to the development of cancer can its elimination or reduction affect the incidence of the 

190. See OTA Report at 138; OSHA Carcinogen Policy at 5038. 

191 . See the discussion of OSHA's health standard for asbestos, OSHA Draft at 125-32; M. 
Shapo, A Nation of Guinea Pigs 191 et seq. (1979). 


is to minimize the effect of confounding variables so that any differences in 
cancer rates can be correlated with chemical exposure. '^^ 

The reliability of the conclusions from either cohort or case-control studies 
depends on the closeness of the match between the compared groups, the accuracy 
of exposure and disease incidence data, and the statistical limits of detecting 
differences in disease incidences between groups. The case-control format has 
particular appeal for the study of possible chemical carcinogens because the 
expected incidence of most cancers is low. The probative weight of findings 
from such a study depends on whether the investigator has identified a control 
population that closely matches the cases. Selecting matched controls requires 
informed guesses about the factors that could affect disease incidence, to which 
individual exposure may often be impossible to verify. '^^ 

Epidemiologists confront three main problems: detection of disease fre- 
quencies; recordation of exposure histories; and discernment of linkages between 
them. The North American data base lags well behind modern needs to describe 
the health effects of environmental agents. The decentralized, predominantly 
private U.S. health care system does facilitate uniform data gathering as readily 
as do state systems. ^^"^ The classical uncertainties of disease recording — inac- 
curacies in mortality and morbidity classification — are thus exacerbated by the 
lack of a uniform national reporting system. Data gathering is also impeded by 
the mobility of the U.S. population, which produces frequent changes in dietary 
and exposure patterns and discontinuities in medical records. The probable costs 
of more accurate and comprehensive health records, furthermore, are very high, 
particularly if this requires a large fraction of deaths to be followed by autopsy. '^^ 

192. For example, the case-control study of the relationship between saccharin consumption 
and bladder cancer first identified a population of individuals with bladder cancer. The cases were 
matched with a group of individuals with similar characteristics who did not have bladder cancer. 
Saccharin consumption in the two groups was determined, often retrospectively, using a sophisticated 
interview process, and then compared. A large case control study of artificial sweeteners and bladder 
cancer has concluded that the general population who reported having used artificial sweeteners 
were at no higher risk of bladder cancer than those who did not. The study, however, did detect an 
increased rate of bladder cancer among certain subgroups studied. Hoover and Strasser, Artificial 
Sweeteners and Human Bladder Cancer, 1980 Lancet 837. An apparent association between coffee 
consumption and pancreatic cancer was also demonstrated in a case-control interview study. The 
coffee drinking habits of 369 patients with pancreatic cancer were compared with those of more 
than 600 controls, who in this study were patients hospitalized for other reasons. See B. MacMahon, 
304 N.E. J. of Medicine 63, 633 (1981). 

193. See OTA Report at 138. 

194. Id. at 15, 142-45. 

195. In addition to these data limitations, a new problem has recently emerged, that of privacy 
limitations. The thorough disease and exposure profiles that are needed to evaluate environmental 
hazards require monitoring of identifiable individuals to an extent unprecedented in American history. 
This problem is both technological and political. Its technological side involves the problem of data 
security balanced against accessibility; its political aspect involves a choice between personal privacy 
and reliable assessment of risks to public health. See Privacy Protection Study Commission, Personal 
Privacy in an Information Society, (GPO 1 977). Although OSHA carcinogen policy requires extensive 
medical surveillance of employees and provides for a detailed completion of employee medical 
records, there is no discussion in the OSHA rule of the privacy issues related to the gathering of 
this information. See OSHA Carcinogen Policy at 5230-34. 


Finally, the number of unidentified carcinogens to which individuals are exposed 
in their daily lives practically defies attempts to control for their effects. 

c. Power and limitations of epidemiological studies. Roughly two dozen 
chemicals have been identified as carcinogens in epidemiological studies. The 
most appealing feature of such studies is the obvious one: they involve human 
populations. Using data from human experience eliminates the uncertainties that 
accompany extrapolation of animal test results to humans. '^^ Epidemiological 
studies can also incorporate, albeit at substantial expense, very large populations, 
which are not practical in animal experiments.'^^ A large population extends the 
lower range of detectable effects and enhances the statistical significance of 
observed associations. 

The successful demonstration that cigarette smoking is associated with lung 
cancer illustrates the power as well as the significant limitations of epidemiology. 
Suggestions in the late 1940's that cigarette smoking might cause lung cancer 
triggered attempts by both experimental researchers and epidemiologists to verify 
the hypothesis. But even extended laboratory efforts were largely unsuccessful 
in demonstrating that animals exposed to cigarette smoke developed lung cancer. 
By contrast, epidemiologists have been able to mount a strong case that cigarette 
smoking causes lung cancer in humans. '^^ 

Early studies compared the incidence of lung cancer in smoking and non- 
smoking human populations, generally agreeing on three observations: (1) The 
incidence of lung cancer among heavy smokers was eight to ten-fold higher than 
among non-smokers. (2) Among groups with differing smoking patterns, in- 
creased exposure to cigarette smoke was correlated with lung cancer incidence. 
(3) The differences in cancer rates between men and women paralleled com- 
parable differences in the proportions who smoked. '^^ Having demonstrated a 
strong association between lung cancer and cigarette smoking, epidemiologists 
had to interpret these findings. Their conclusion that smoking causes cancer 
rested on a two-branched rationale: First, the observations were consistent with 
the expectation that cigarette smoke was a carcinogenic agent. Second, based 

196. In addition, the environments in which chemicals have been associated with human 
cancer are more hkely than the controlled setting in which animal experiments are conducted to 
resemble other human environments. 

197. B. MdcMdhon. Epidemiology's Strengths, Weaknesses, in Environmental Health Issues, 
30 National Academy of Services, News Report, No. 4. p. 5 (April 1980). 

198. See generally U.S. Public Health Service, Smoking and Health: Report of the Advisory 
Committee to the Surgeon General. Public Health Service Publication No. 1 103 (Department of 
HEW 1964). 

199. The Health Consequences of Smoking: A Public Health Review, Public Health Service 
Report p. 9. (Department of HEW 1967). Epidemiologists have demonstrated an increase in the 
incidence of lung cancer among women that follows, with the expected lag time of between twenty 
and twenty-five years, a sharp increase in cigarette smoking among women. These observations 
have emerged repeatedly in independent studies. The Health Consequences of Smoking for Women, 
Public Health Service Report, pp. 1 15-1 16 (Department of HEW 1980). 


on the available data, other explanations for these observations did not seem 
plausible. ^^ 

This description illuminates several features of epidemiological investiga- 
tion. First, the finding that smoking was associated with lung cancer was only 
the first, and easiest, step. Determining whether smoking causes cancer required 
considerable research to eliminate other proposed explanations by showing that 
their predicted consequences were inconsistent with the available observations. 
Experiments were repeated to confirm the reliability of the observations and 
others were conducted to eliminate alternative explanations, requiring consid- 
erable time and expense. ^^' More important, these efforts were necessary to 
detect the effects of an agent that is now blamed for as many as 30 percent of 
all cancers and a higher percentage of lung cancers; even the best human studies 
are not capable of discerning modest increases in disease rates caused by ex- 
ogenous factors — increases of the magnitude likely to be associated with exposure 
to toxic chemicals. ^^^ 

Accordingly, while epidemiological investigations have illuminated the role 
of environmental factors in some human diseases, they have played a limited 
role in identifying health effects of specific chemicals. Their expense alone would 
preclude routine use of epidemiological studies as a screen for carcinogenic 
effects. Moreover, epidemiologists can evaluate only those substances to which 
humans are already exposed; ethical constraints generally preclude nonthera- 
peutic experiments with human subjects. ^^^ In consequence, the data used in 
epidemiological studies are often poor. Epidemiologists do not conduct "ex- 
periments"; instead, they take human circumstances as they find them and 

200. The evaluation process that supports the latter conclusion in the case of smoking is 
illustrated by the following dialetic. It was once suggested that other airborne chemicals were inhaled 
by humans and could have been causes of lung cancer. This explanation was undermined, however, 
by the finding that men and women, equally exposed to general environmental influences, displayed 
dramatically different rates of lung cancer. Another alternative, taking into account the differential 
rates between the sexes, then suggested that men were more susceptible than women and for this 
reason exhibited higher rates of lung cancer. But this hypothesis did not account for the later 
observation that the increase in lung cancer among women paralleled in time the rise in the proportion 
of women who smoked. Thus, despite energetic attempts to suggested other theories, the most 
convincing, straight-forward explanation for all of the observations remained that smoking causes 
lung cancer. 

201. It is quite possible that altSmative explanations for an observed association between 
chemical exposure and cancer cannot be convincingly eliminated and that the causal character of 
the association will remain, after many studies, only one of several hypotheses. 

202. See OTA Report at 67; OSHA Carcinogen Policy at 5049. 

203. See generally, J. Katz, A. Capron, and E. Glass, Experimentation with Human Beings 
(1972). Compliance with Department of Health and Human Services regulations for human studies 
is required to obtain federal research funds. These regulations, currently under revision, are authorized 
by 5 U.S.C. § 301 (1976); 42 U.S.C. §§ 2891-93(a), and appear in 45 C.F.R. Part 46 (1979). It 
is sometimes possible to design prospective epidemiological studies that will yield information about 
health effects associated with a new substance as human exposure continues, but this information 
cannot aid initial decisionmaking. An example of a prospective study on smoking is E. C. Hammond, 
Smoking in Relation to the Deaths of One Million Men and Women, 19 National Cancer Institute 
Monograph 127 (1966). 


attempt to select from recorded data the information required for a well-designed 
study of carcinogenicity. Most available data, e.g., from hospital or employer 
records or death certificates, are at best an imperfect approximation of actual 
experience. Their deficiencies are particularly evident in cancer studies, which 
often try to reconstruct conditions that existed many years earlier at the time of 
initial exposure to a chemical. ^^ 

d. Statistical power. Probably the most important and yet least understood 
limitation of epidemiological studies is their inability to reveal, or to negate, 
any but large associations between exposure and disease. This limitation of 
epidemiological studies is a function of their statistical power. 

The hypothesis underlying such studies is that exposure to a substance that 
causes cancer should increase its incidence. Testing this hypothesis requires 
comparing the cancer rates in at least two populations. The protocol for making 
such a comparison is to locate two groups that differ in their exposure to the 
chemical but are closely matched in all other potentially significant variables. 
Ordinarily, however, some individuals in the unexposed group, or "controls," 
will have developed the type of cancer suspected to be caused by the chemical. 
Only if the incidence of cancer in the exposed group stands out above the 
"background" incidence in the unexposed group will an association be detected. 

Once the investigator has measured the cancer incidence in the two groups, 
he must decide whether the incidences in fact differ. Successive independent 
measurements of disease incidence in two similar groups would not be expected 
to be identical; rather, they should cluster around a mean value. This mean value 
is lilcely to approximate the true incidence of cancer in the population of which 
the measured groups are representative. The crucial question in comparing two 
groups is whether divergent measurements reflect different incidences or simply 
chance fluctuations about the same mean. Some sets of measurements are better 
able in a probabilistic sense than others to reveal whether different measured 
incidence values represent real differences. This property of discrimination, more 
precisely the likelihood of obtaining a statistically significant measure, is termed 
statistical power. -^^^ In a study that possesses high statistical power, differences 
in measured cancer incidence between the exposed and unexposed groups can 
be confidently interpreted to mean that the true incidences actually differ. In 
general, studies that reveal only a relatively low excess incidence of cancer in 

204. One objective of a carcinogenesis study is to demonstrate, if possible, that levels of 
exposure to a chemical correlate with the incidence of the disease. But direct measurement of past 
exposures is rarely feasible and methods for approximating exposure levels are usually crude. For 
example, investigators interested in the effect of smoking cannot measure the quantities of carcinogens 
in cigarette smoke or even the volume of cigarettes smoked in a day many years earlier. They are 
reduced to relying on the study participants' memories as to how many cigarettes a day they have 
smoked over their lifetimes. Lave, The Strategy of Social Regulation 92-97 (1981). Authors of a 
case-control interview study on coffee asked patients to estimate the number of cups of coffee per 
day they had been drinking before their present illness. See B. MacMahon, note 206 supra. 

205. Statistical power is a function of the ratio of the magnitude of the reported excess cancer 
risk to the spread of those values around the mean and of the size of the population studied. See C. 
Hand, Estimating Cancer Risks from Low Doses of Ionizing Radiation, 209 Science 1197, 1197, 


the exposed group have low power and are less capable of discerning small 
actual differences between two groups. Because the increases in risk expected 
from low-dose exposure to carcinogens are usually small, studies of such ex- 
posure very often lack statistical power. 

Increasing the size of the two groups can enhance the statistical power of 
a study, but large populations are rarely feasible. For example, to detect a very 
low increased incidence of cancer, e.g., fewer than 10 cases per million, over 
a reasonably high background rate, e.g., 1900 cases per million, a study would 
require a population of 100 million. ^^^ Such a study, obviously, could never be 
conducted. And even when the populations are large enough to allow confident 
measurement of differences in cancer incidence, a high background incidence 
in the controls may make it impossible to determine whether the differences are 
real. Accordingly, epidemiological studies are rarely capable of determining 
whether exposure to a chemical is associated with an excess risk of cancer. 

This lack of statistical power also weakens the significance of negative 
epidemiological findings. An example will illustrate this point. Suppose that 
investigators desire to determine whether exposure to chromium causes brain 
cancer. A careful study might involve examination of the death certificates of 
1 ,000 chromium plant workers to determine their incidence of brain cancer and 
comparison with the incidence of brain cancer among 1000 workers of the same 
sex, race, age, habits, etc., in another industry. The failure of such a study to 
find an excess risk of brain cancer among chromium workers would not exculpate 
chromium as a possible carcinogen. A study in a population of only 1000 would 
not detect even one excess brain cancer if the true risk from chromium were, 
say, 1 in 10,000 — not a notably low risk for ambient exposure to a carcinogen. 
Furthermore, even if the populations were large enough to detect such a risk, a 
negative result would not warrant a conclusion that chromium posed no risk 
because the study would probably lack the power to resolve the expected small 
difference in measured brain cancer incidence between exposed and unexposed 
groups. ^^^ Uncertainties in recordation of the disease and of exposure levels 

206. /i/. at 1198. 

207. Since the cancer incidence attributable to chromium is so low, the study could not 
confidently distinguish a small excess incidence attributable to chromium exposure from a fluctuation 
in the background incidence of brain cancer expected in the control population. Most epidemiological 
studies are limited to demonstrating excess risks which are relatively high with respect to the incidence 
exhibited by the general population. But see R. W. Morgan, Some Notes on Epidemiology 198 
(AIHC 1980). Morgan advocates increased reliance on large epidemiological experiments to evaluate 
low magnitude health risks. However, the sample sizes he visualizes as providing sufficient power 
to detect relatively low (10 to 20%) excess risks under several different conditions range from tens 
of thousands to hundreds of thousands of cases. 

According to Dr. Robert Hoover, an epidemiologist at the National Cancer Institute, the 
lowest relative risk of cancer accepted as having been established by an epidemiological study was 
30-40%. OSHA Carcinogen Policy at 5051. Dr. Richard Bates claimed that "few epidemiological 
studies of occupational cancer have been of adequate sensitivity to detect anything smaller than a 
50% excess risk." Id. at 5049. 


would undermine negative, just as positive, findings. ^^^ It therefore would be 
inappropriate to conclude, from a study that failed to find an excess of brain 
cancer among chromium workers, that chromium does not cause brain cancer 
or that exposure to it is "safe." 

Most chemical carcinogens are expected to cause such small excess risks 
that they defy detection by standard epidemiological methods. On the other hand, 
human studies could detect any large excess of disease associated with exposure 
to a chemical. Negative findings can thus help define an upper limit of excess 
risk posed by a substance. The smallest excess risk that a well-designed negative 
study would be capable of measuring can be taken as one approximation of the 
upper level of risk that the chemical poses. ^^^ 

e. Uses of epidemiology in regulatory decision making. Of the four agen- 
cies studied here, only OSHA's decisions have been influenced decisively by 
epidemiological evidence. The industrial workplace is one of the few settings 
in which human exposures to toxic chemicals have been high enough, prolonged, 
and sufficiently documented, to permit serious investigation of their health ef- 
fects. The following discussion illustrates the use of human data in evaluating 
the risks of one occupational exposure. Thereafter, the still-unsuccessful epi- 
demiological efforts to define the risks of saccharin are described. 

The difficulties of conducting and interpreting epidemiological studies are 
illustrated by a prospective study of steel workers, which became the main basis 
for OSHA's reduction of the permissible exposure limits for coke-oven work- 
ers. ^'° The investigators selected a sample of workers employed on a particular 
date. One problem with this sample was that the majority of workers had been 
employed for more than one year; it thus excluded most workers with allergic, 
'cardiovascular, or respiratory conditions that might make them sensitive to chem- 
icals in the steelmaking environment.^" It became apparent later that steel- 
workers were significantly healthier than the general population of their same 
age, sex, and race — a phenomenon that complicates many occupational stud- 
ies. ■^''^ As in many other occupational studies, the investigators failed to obtain 
information about workers' initial health status, their personal habits (especially 

208. For example, a complete study of the effects of chromium would include the causes of 
deaths of chromium workers who left the plant as well as those who completed their careers there. 
Some workers may have left precisely because they became sick. While information of this kind is 
usually hard to obtain, its absence weakens the significance of any negative findings. 

209. OSHA Carcinogen Policy at 5057-59. 

210. Lave, The Strategy of Social Regulation 93-98 (1981). 

211. /c/. at 94. An alternative design might have focused on workers hired after the beginning 
of the study, with special attention given to those who left employment or experienced health 
difficulties, but such a design would reduce the sample size and the number of years of exposure. 

212. Id. It would be important to know how much healthier would these workers be expected 
to be in the absence of exposures to toxic substances and higher accident rates. A 5% measured 
increase in life expectancy for workers over that of general population might nevertheless conceal 
a large occupational risk because the improvement in health should have been much greater in the 
absence of occupational exposures. See Mazumdar and Redmond, Evaluating Dose Response Re- 
lationships Using Epidemiological Data on Occupational Subgroups. OSHA Carcinogen Policy at 
5053 in Breslon and Whittemore, eds.. Energy and Health (1979). 


cigarette smoking), their family disease history, or their previous employment 
history, any of which could affect the health patterns observed. ^'^ 

Early in the study the investigators found that coke oven workers exhibited 
much higher cancer mortality rates than the population generally or other steel- 
workers. Since only a small proportion of steelworkers are employed at coke 
ovens, however, their sample was small. Furthermore, as concern focused on 
lung cancer, the failure to collect data on cigarette smoking became a serious 
deficiency. The suspicion that the excess rate of lung cancer could be due to 
heavier smoking by coke oven workers was finally put to rest because of the 
extraordinarily high incidence of lung cancer among these workers, whose rates 
differ with the amount of time worked and the type of job performed — factors 
that reflect levels of exposure. 

While the study thus demonstrated that exposure to coke oven chemicals 
sharply increases the risk of lung cancer, it did not provide sufficient information 
to permit confident estimates of the dose-response relationship. If the number 
of cancers among the workers were known, if work histories provided a good 
estimate of the dose, and if current measurements of coal tar pitch volatiles 
characterized the relevant set of coke ovens for the relevant period of time, the 
researchers would have had the needed data.^''* But all of these variables were 
obviously subject to large errors of measurement, and only a few data points 
could be used in the estimation. Several extrapolation models thus fit the data, 
but they produced widely different estimates of the amount of lung cancer that 
could be expected if exposure were reduced. ^'^ 

3. Animal Experiments. 

a. Background. While human studies have demonstrated the carcinogen- 
icity of approximately two dozen chemicals,^ '^ their significant limitations di- 
minish their value in regulatory decisionmaking. Toxicology has provided an 
alternative in the form of experiments in laboratory animals. These experiments 
comprise the primary data for regulating toxic chemicals, and there is both 
theoretical and empirical support for their use in assessing human risk. Most 
human carcinogens cause cancer in several mammalian species; it is thus possible 
to derive evidence of a chemical's potential human carcinogenicity from tests 
in other species. Even if the correlation between human and animal responses 
to carcinogens were weaker, however, regulators would be induced to seek 
alternatives to epidemiological evidence. To evaluate the risk of chemicals before 
human exposure occurs, agencies must rely on testing in surrogates. Animal 
experiments have the conspicuous advantage of providing a source of data from 
expendable species. 

213. Id. For example, asbestos workers who smoke are at many time the risk of non-smoking 

214. Lave, supra note 210. 

215. Id. at 96. 

216. OTA Report at 141. 


Toxicologists agree that, although not ideal, the best practical test for iden- 
tifying causes of human cancer is the long-term bioassay in rodents.^ '^ Though 
their details still occasion scientific debate, these tests generally follow a well- 
accepted pattern: ^'^ A small population of animals (e.g., rarely more than 400), 
which have been maintained under similar conditions in a controlled laboratory 
setting, is divided into groups equally distributed by sex, e.g., 50 animals of 
each sex per group. One and usually more of these groups is regularly exposed 
to the test chemical for the major part of their lifespan. A control group of similar 
size is left untreated. All animals are carefully observed for the appearance of 
tumors during the study and, at the conclusion, are sacrificed and examined 
pathologically. The incidence(s) of tumors observed in the exposed group(s) is 
compared to the incidence of tumors in the controls. Because the groups are 
otherwise theoretically identical, any increase in tumor incidence in the exposed 
animals presumably can be attributed to the test chemical. 

Although animal bioassays are straightforward in concept, they are difficult 
to design, execute and evaluate. The standard rodent bioassay requires three to 
five years to complete and demands coordinated efforts by scientists and tech- 
nicians who possess skills ranging from routine animal care to pathological 
diagnosis. The cost of evaluating one chemical in a single rodent species now 
substantially exceeds $500,000.^'^ Each element in the design of an experiment 
calls for scientific judgment, which later may provide the basis for argument 
when a regulatory agency seeks to interpret the results. The following discussion 
highlights common sources of disagreement. 

The critical design parameters of animal carcinogenicity tests are the species 
used, population size, and the dose levels of the test chemicals. Elaborate though 
they are, current tests reflect major compromises of the theoretical ideal on each 

b. Choice of species. The suitability of an animal species as a model for 
evaluating human health effects in theory should depend on the similarity of its 
reactions to human reactions. If the sole criterion were biological similarity, 
primates would probably be the preferred experimental species. Primates are 
used in some testing programs, ^^^ but they have not been seriously considered 

217. General Criteria for Assessing the Evidence of Carcinogenicity of Chemical Substances , 
Report of the Subcommittee on Environmental Carcinogenesis of the National Cancer Advisory 
Board, 58 J. Nat'l. Cancer Inst. 461-465 (1977). 

218. EPA, Interim Procedures and Guidelines for Assessing Health Risk and Economic Im- 
pacts of Suspected Carcinogens, 41 Fed. Reg. 21402 (1976); J. M. Sontag, N. P. Page, and S. 
Saffiotti. Guidelines for Carcinogenesis Bioassays in Small Rodents, DHEW Pub. No. 76-801 (1976) 
[hereafter NCI Bioassay Guidelines]. 

219. EPA, Support Document for Test Data Development Standards for Chronic Health 
Effects, under Section 4 of the Toxic Substances Control Act ( 1 979) [hereafter EPA Test Development 

220. Primates are commonly used in vaccine testing, and occasionally in testing food additives. 
Office of Technology Assessment, Cancer Testing Technology and Saccharin, Appendix I, pp. 79, 
81 (1977) (monkeys were used in experiments to test saccharin) [hereafter OTA Saccharin Report]. 
Primates are expensive and difficult to handle in the laboratory. Successful breeding colonies are 
few and repeated raids on wild populations raise both ecological and ethical issues. 


as a model for carcinogenicity testing. Primates are costly to obtain and house, 
and their long life span — comparable to man's — impractically prolongs the period 
needed to observe effects of chronic exposures. ^^' Cost and ease of maintenance 
are thus powerful incentives to select small, short-lived mammals, and rats and 
mice have accordingly been selected as the primary species for carcinogenicity 
testing. These rodents can be conveniently maintained at moderate expense and 
their short life-spans (2-A years) permit discovery of tumors that only become 
manifest late in life. Furthermore, they are available in several well-characterized 
genetic strains whose biology is well-documented. ^^^ 

Geneticists have created several inbred strains of rats and mice which are 
genetically homogeneous. In this respect, these animals differ from the human 
population, ^^^ but inbred strains are nonetheless widely used because their very 
homogeneity facilitates interpretation of test results. The variability in the re- 
sponses of outbred animals can obscure differences between control and exposed 
groups; the responses of inbred strains are more easily predicted. ^^"^ 

To be a useful experimental model a rodent strain must be capable of 
contracting cancer. Otherwise, a failure to detect tumors could reveal nothing 
about the test chemical. Theoretically, administration of a recognized carcinogen 
to one group, a so-called positive control, should be part of any carcinogenicity 
test. Routine experiments rarely include positive controls, however, for past 
observation has established that the strains used in testing are susceptible to 
cancer. -^^^ 

Although most carcinogens cause cancer in several mammalian species, 
few do so in all species and fewer still induce tumors in the same organs in each 
vulnerable species. The lack of uniform response across species undercuts the 
predictive value of single experiments in two ways. First, a positive result in a 
single species is less reliable as a forecast of carcinogenicity in other species. 
More significantly, a negative response in one species provides weak evidence 
that the chemical will fail to induce tumors in other species. A conclusion that 
a chemical is not carcinogenic usually requires negative findings in two sexes 
of two mammalian species. ^^^ By contrast, a single finding that a chemical 
increased the incidence of tumors in one sex of one species is generally accepted 

221. There is evidence that the latency period for tumor induction is related to the lifespan 
of a species. The choice of a long-lived species thus prolongs the time required to conduct a single 

222. Largely fortuitously, rats have been found to possess some striking physiological sim- 
ilarities to humans which particularly suit them for use as experimental models for assessing human 

223. See NAS/NRC, Principles for Evaluating Chemicals in the Environment 1 15-133 (1975). 

224. NCI Bioassay Guidelines. 

225. M. Schneiderman and C. Brown, Estimating Cancer Risks to a Population, 22 Envi- 
ronmental Health Perspectives 115, 118 (1978). 

226. EPA Test Development Standards at II-9. The two-species criterion is a practical measure, 
which does not assure that no agent which is carcinogenic for man will slip through. Arsenic is an 
example, albeit a rare one, of a chemical that fails to induce tumois in animal tests but is a known 
human carcinogen. V. Bencko, Carcinogenic, Teratogenic and Mutagenic Effects of Arsenic, 19 
Environmental Health Perspectives 179-82 (1977). 


as indicating a potential carcinogenic hazard to humans. ^^^ Evidence of carcin- 
ogenicity in a second sex or species reduces concern that some deficiency of the 
first test species produced the increased tumor rate.^^^ Data from positive ex- 
periments in two species can also facilitate quantitative assessments of the risk 
of cancer for humans. ^^'^ Because even species that respond to a carcinogen vary 
in their sensitivity, an agency will usually select data from the more sensitive 
species to estimate human risk.^^° 

Most animal strains exhibit some tumors even without deliberate exposure 
to a known carcinogen."^' A high spontaneous tumor rate can obstruct analysis 
of an experiment. Just as in human studies, as a species' background incidence 
of tumors rises, the incidence of induced tumors must also rise to be detectable. ^^^ 
If there were no practical limits on experiment size, a high spontaneous tumor 
level could be accommodated by increasing the number of animals on test. But 
the number of animals now used in testing is already near the practical maximum; 
therefore a high spontaneous tumor rate is a serious disadvantage in a test 
species. ^^^ 

c. Number of animals. The number of animals used is perhaps the major 
limitation of routine bioassays. Small populations limit experiment sensitivity 
and dictate the use of high doses of test chemicals. ^'''^ Though confusing to many 
laymen, the relationship between population size and test sensitivity is straight- 
forward and fundamental. 

It is obvious that exposure of ten animals to a chemical that is capable of 
causing cancer will frequently not detect an effect if the chemical induces tumors 
in only one of every hundred exposed individuals. Even if 100 animals were 

227. IRLG Scientific Principles at 263; OSHA Carcinogen Policy at 51 18. See also CPSC, 
Interim Policy and Procedure for Classifying. Evaluating and Regulating Carcinogens in Consumer 
Products, 43 Fed. Reg. 25658. 25663 (1978) withdrawn. 44 Fed. Reg. 23821 (1979). 

228. See, e.g., EPA Test Development Standards at II-9. In dealing with politically sensitive 
chemicals agencies may see fit to fortify themselves with far more than two positive studies before 
acting. For example. FDA received the first study showing a risk of bladder cancer in second 
generation rats exposed to saccharin in 1972. The agency however did not announce its intention 
to ban saccharin until 1977 after receipt of several more studies, excluding the Canadian bioassay 
which had prompted a ban of saccharin by the Canadian government only one month prior to the 
FDA announcement of its contemplated ban. OTA Saccharin Report at 50-79. (OTA cites 10 studies 
of varying quality done between 1948 and 1977.) 

229. Repeated experiments and large numbers of animals are important statistical parameters 
which tend to increase confidence in the experimental conclusions. 

230. EPA Test Development Standards at II- 10; M. Schneiderman and C. Brown, Estimating 
Cancer Risks to a Population, 22 Environmental Health Perspectives 1 1 5, 1 1 7 ( 1978); OTA Saccharin 
Report at 87. 

231. In some cases, the occurrence of spontaneous tumors by animals in a cancer test can 
provide confirmation of the capability of the strain to develop tumors. Observation of spontaneous 
tumors provides evidence of a strain's general capacity to develop tumors, but the induction of 
tumors by a chemical similar to the test chemical is better evidence that the strain has the capacity 
to respond to the test chemical if it is carcinogenic. 

232. See Table I, infra. 

233. EPA Test Development Standards at 11-19 and 20. 

234. Id. at 17-19; OTA Saccharin Report, Appendix 1-85. 


used, it is not certain that one, and only one, case of cancer would appear. A 
cancer incidence of 1 in 100 is a statistical expression, which means that in 
repeated tests exposing 100 animals, the number of afflicted animals may vary 
but will be distributed around a mean value of one. Therefore, if the incidence 
of tumor induction were 1 in 100, it would be expected that in a series of tests 
each using 100 animals, some tests would reveal no afflicted animals, most 
would reveal one, while a few would reveal two or three or even more. When 
the possibility of observing no afflicted animals — and thus failing to detect the 
chemical's effect is great, a test system is inadequate. Increasing the size of the 
groups is one way of increasing the test sensitivity. 

Table I. 

Incidence of Tumors in Exposed Groups Required for Significance 
(/7 = 0.05) Depending on Test Group Size and Incidence of Tumors in 


Incidence of Tumors 

in Controls {%) 
(spontaneous tumors) 



No. of Animals 
Per Group** 

































*Calculations based upon tabulationsof Mainland and Murray (1952) and presented by Page (1977a). 
** Exposed and control groups of same size. 
Source: EPA Test Development Standards at 11-19. 

Table 1 illustrates the relationship among test sensitivity, group size, and 
background incidence of tumors. Where the spontaneous or background incidence 
is zero percent, the minimum incidence of afflicted animals (tumors) considered 
experimentally significant in a group of 100 test animals is 6 percent; in a group 
of 50 animals, it is 12 percent. The spontaneous occurrence of tumors in the 
test species increases the minimum incidence that can be detected experimentally. 
Given a background incidence of 10 percent, groups, of 50 animals are the smallest 
in which a reasonable incidence of tumors (less than 40 percent) can be distin- 
guished from chance fluctuation. Thus, a test using groups of that size can detect 
only a very high incidence of tumors. Even the most potent human carcinogens, 
however, are not encountered at levels that produce tumors in 12 percent of the 


Practical constraints discourage continued expansion of the size of groups 
to increase test sensitivity.^'*'' Increasing the sensitivity of the standard single- 
species experiment by an order of magnitude would require enlarging each group 
to 500 animals of each sex, a total of 3,000 animals for a two-dose experiment. 
The expense of such experiments, not to mention the lack of trained pathologists 
to interpret them, precludes routine testing on this scale. 

d. Dose levels. The practical limits on the size of animal experiments 
have led to the use of high test doses, which are necessary to produce a detectable 
incidence of tumors. Experimenters customarily use the highest dose the animals 
can tolerate without experiencing other unacceptable adverse health effects. ^^^ 
Higher doses may overload the animals' system and distort their metabolism of 
the test chemical; even the "maximum tolerated dose" (MTD) may produce 
other adverse effects in the animals. More important, high test doses complicate 
interpretation because they generally exceed, sometimes by several orders of 
magnitude, the levels at which humans are likely to be exposed to the test 
chemical. Despite these difficulties, toxicologists generally support the use of 
high experimental doses to determine whether a substance is capable of causing 
cancer.-'*'' This practice has, however, often occasioned public misunderstanding 
of test results and sometimes evoked barbed criticism of agency reliance on 
animal data.^"*^ 

A desired product of any animal bioassay is a positive correlation between 
increasing dose and response which supports the inference that the chemical 
caused the response. In addition, data demonstrating effects at several dose levels 
can be helpful in estimating the degree of human risk. Accordingly, it is now 
common practice to administer two or even three dose levels of the test chemical, 
usually the MTD and at least one lower dose. However, positive responses from 
a broad range of doses are not required to demonstrate a chemical's carcino- 
genicity, and routine bioassays frequently fail to produce a dose-response curve. 
A test's lack of sensitivity may restrict the dose range at which a response can 

235. A small increase in size would not be meaningful. Even a ten-fold increase would not 
produce the sensitivity necessary to detect the effects of likely levels of human exposure to most 
carcinogens, but it would severely strain practical test limits. Using five hundred instead of fifty 
rats of each sex per group, a three-dose experiment would require 4000 rats. The care, feeding, and 
daily observation of this many rats is no small task, but is only the beginning of the work. After 
each animal dies it must be autopsied, and its organs prepared for histological examination. Then 
a series of slides from each of the organs of each animal must be individually evaluated for evidence 
of tumor formation. Repetition of the experiment in a second species would double the number of 
animals to 8000. 

236. There is no general agreement on the parameters for a maximum dose. According to 
NCI, the two important criteria for determining the maximum tolerated dose are that the dose neither 
shorten the life span of the animal nor retard weight increase by more than 10 percent compared 
with controls. NCI Bioassay Guidelines OTA Reports at 123-26. EPA on the other hand has defined 
its high dose level to be the "maximum dose level that can be administered for the duration of the 
test period with demonstrable but only slight toxicity in test animals, and no substantial reduction 
in longevity due to effects other than tumors." EPA Test Development Standards II 21-22. 

237. OTA Report at 125-26. 

238. OTA Saccharin Report at 


be reliably detected. Furthermore, the selection of appropriate doses on the basis 
of sub-chronic tests is not foolproof. For example, the highest dose administered 
may turn out to have unacceptable adverse effects, or the lowest dose may 
produce no response at all. 

A "no effect" level is the dose level of a chemical, known to be toxic at 
high doses, at which — and thus presumably below which — adverse effects are 
not observed. No-effect levels can be determined for many non-carcinogens and 
are commonly used by regulators in setting exposure levels that are presumed 
safe. But routine animal bioassays are not capable of demonstrating a no-effect 
level for a carcinogen. First, their insensitivity makes them incapable of ruling 
out the possibility of tumors in a small percentage of exposed individuals. ^-^^ In 
addition, assumptions about the mechanism of carcinogenesis are consistent with 
the occurrence of tumors at extremely low doses. Thus there currently is no 
practical method for determining no-effect levels for carcinogens experimen- 
tally."^"*^ The most any negative bioassay can convincingly demonstrate is the 
maximum risk of cancer that a chemical may pose, i.e., the incidence that could 
escape detection. 

e. Route and duration of exposure. The route by which an animal is 
exposed to a carcinogen may not only influence the sites at which tumors develop; 
it can determine whether tumors occur at all. Common experimental routes of 
exposure include ingestion (in food or water), intubation (force-feeding through 
a tube), inhalation, and direct application to the skin (dermal exposure). The 
route of exposure in a bioassay is chosen by convenience and, in studies con- 
ducted specifically for use by regulators, by analogy to probable routes of human 
exposure. Ingestion is by far the most common route of administration, but 
feeding studies require regular monitoring of the animals' intake of food and 
water to determine the doses to which the animals were exposed. 

Exposure to the test chemical is usually continuous from time of weaning 
to some specified date near the end of the animals' lives, usually 24 to 30 months 
from the commencement of the study. Lifetime exposure maximizes the total 
dose and theoretically compensates for the insensitivity of the test.""*' The period 
of exposure is often extended by beginning treatment in utero, exposing pregnant 
females and then continuing administration to the offspring throughout their 
lives. ^"^^ Occasionally, exposure is terminated prior to the end of the animals' 
lives to permit observation of any regression of reversible growths. 

f. Conduct and evaluation. A bioassay requires at least two years to 
conduct and several more months to analyze and report, and demands attention 
to detail at every stage. The animals must be maintained in clean surroundings, 
exposed according to pre-determined schedule (usually daily) to the test chemical, 
and carefully handled to avoid interruption of treatment or mixing of the groups. 

239. See Table II, supra. 

240. OTA Report at 125-26; 167-69. 

24 1 . EPA Test Development Standards 11-15. 

242. OTA Saccharin Report at 51-60. Only "two generation" studies have consistently 
showed saccharin to be a bladder carcinogen in rats. 


Each animal should be examined periodically for signs of contagious disease as 
well as for early evidence of tumors. Laboratory personnel must be particularly 
alert to detect dying animals and act promptly to preserve organs and tissues. 
Autolysis, or disintegration of tissue, commences immediately after death and 
destroys the delicate cell structure which enables pathologists to diagnose and 
classify tumors. ^'*^ Furthermore, rats and mice have a tendency to cannibalize 
the remains of recently deceased cagemates. Routine of observation is tedious 
but vital to the reliability of experimental findings. 

At the predetermined date,'^'^'^ all surviving animals are sacrificed, their 
organs are examined visually,-'^'' and selected tissues are then prepared for mi- 
croscopic examination. In elaborate studies, slides will be prepared from between 
30 and 40 tissues from each animal. ^"^^ This histopathology is a costly and critical 
stage of the evaluation process. 

The final stage of the study is the diagnosis of any tumors each animal 
experienced and the computation of the incidence of each type of tumor for each 
exposure group. -"^^ Such diagnosis depends primarily on the appearance of tissues 
under a microscope and can be performed only by trained pathologists. Clas- 
sification of observed tumors is rarely unequivocal; it usually calls for judgments 
on which even the best pathologists may disagree. Some study protocols specify 
that slides are to be evaluated by more than one pathologist, and regulatory 
agencies often seek the judgment of several experts in interpreting the results of 
controversial tests. '^'^*^ 

243. In an experiment using only 50 annuals per dose groups, the loss of even a few animals 
can severely impede determination of tumor incidence. As the experiment progresses, information 
accumulates in the physical states of individual animals. It therefore becomes increasingly important 
to take the precautions necessary to avoid losing animals. EPA may not accept studies with losses 
greater than 5 percent per group. EPA Test Development Standards 11-34. 

244. E.g. , EPA requires that substance be administered for 24 months. EPA Test Development 
Standards 11-20. 

245. Id. at 29-30. 

246. The number of tissues to be examined is a matter of debate. Id. at 11-41 . EPA requires 
detailed microscopic examination of 30 to 40 tissues in each animal. Id. at II-5. 

247. /t/. at 11-41. 

248. An example of the lack of precision in the diagnosis of the presence and type of cancer 
in an animal carcinogenicity came to light when FDA abruptly withdrew its proposal to phase out 
the use of nitrite on the ground that the tissue slides from the animal test ostensibly showing nitrite 
to be a carcinogen had been incorrectly interpreted. The agency's original decision to phase out the 
use of nitrite had been based on a single well-designed study conducted by a respected researcher. 
Dr. Paul Newbome at MIT. But a review of the slides from the study by a panel, which included 
scientists from three federal agencies and other qualified non-government pathologists, uncovered 
significant disagreements with the author's report. In the reviewer's opinion Newbome had misdi- 
agnosed a tumor type that was known not to be related to nitrite administration. The reviewers also 
disagreed with Newbome's diagnosis of other lesions as precursers of cancer. Calculations based 
on data from the reclassified slides failed to show any increase in lymphona among animals treated 
with nitrite compared to controls. 

The expenditure of time and resources necessary to resolve this particular controversy underscore 
the fact that histopathological diagnosis of tissue slides is far from a routine operation. Disagreement 
among experts is not uncommon. This has promoted the suggestion that a panel of pathologists be 
used instead of a single expert to evaluate the slides from animal tests. Implementation of such a 
proposal would further increase the demands of animal toxicity testing on the professional pathological 
resources. See Smith. Nitrites: FDA Beat.s a Surprising Retreat, 209 Science 1 1 10 (1980). 


g. Distinguishing initiators and promoters. Relatively little is known about 
the mode of action of most chemical carcinogens. There is, however, growing 
support for the hypothesis that they can be classed in two broad categories: 
initiators and promoters. -"^'^ According to this hypothesis, initiators interact with 
normal cells to produce cell types that are capable of developing into a tumor. 
These "initiated" cells remain dormant, however, unless they are also exposed 
to a promoter, which can stimulate their development to a mature tumor. ^^^ 
Many chemicals identified as initiators also apparently have promoter properties 
and are thus believed capable of supporting the complete development of tumors, 
i.e., they are complete carcinogens. Other initiators seem to require the collab- 
oration of a promoting agent. Agents that are only promoters are incapable of 
producing tumors in the absence of initiation. 

While this hypothesis of two-stage carcinogenesis has growing support, the 
process is far from understood. If it gains general acceptance, the ability to 
differentiate experimentally between the two types of molecules might become 
important. If a chemical could be reliably characterized as a promoter, infrequent 
human exposure might entail very little risk of cancer. Standard animal exper- 
iments, however, are not now designed to permit such distinctions, and little 
progress has been made in developing tests capable of routinely classifying 
chemicals as promoters or initiators.-''' Furthermore, it is uncertain whether the 
information, even if obtainable, would materially aid decisionmaking about spe- 
cific chemicals.'^''- All animals are constantly being exposed to a vast array of 
chemicals. It is possible that all mammals contain dormant initiated cells whose 
development depends on exposure to promoters. Conversely, the milieu con- 
ducive to promotion may be endogenous and tumor development may await 
exposure to initiators. The prevalence of spontaneous tumors indicates that most 
animal species have either encountered or contain agents with both properties. 
Accordingly, even if a chemical could be classed as a promoter, knowledge of 
the biological state of exposed humans would be needed before this classification 
could be significant for regulation. 

4. Short-term Tests. 

In the last decade scientists have designed a variety of short-term tests to 
reveal information about the potential carcinogenicity of chemicals. The results 

249. See generally, J. L. Marz, Tumor Promoters: Carcinogenesis Gets More Complicated, 
201 Science 515 (1978). 

250. Biochemically, initiators are thought to interact directly to modify the cells' genetic 
material. See Berenblum, Theoretical and Practical Aspects of the Two-Stage Mechanism of Car- 
cinogenesis 25, 26, in Carcinogens: Identification and Mechanism of Action (1978). Promoters 
however are thought to act by other mechanisms, perhaps as cell growth regulators. Bouthwell, 
Biochemical Mechanism of Tumor Promotion 49, in 2 Carcinogenesis, Mechanism of Tumor Pro- 
motion and Cocarcinogenesis (1978). 

251. OTA Report at 66. 

252. A. C. Kolbye, /i U.S. Viewpoint: Legislative and Scientific Aspects of Cancer Preven- 
tion, 9 Preventive Medicine, 267-274 (1980). But see C. Campbell, A Novel Decision Free for 
Regulatory Decisions on Food Chemicals, 57 Toxicology and Applied Pharmacology (1981), for a 
regulatory decision making mechanism which takes into account the classification of a chemical as 
either an initiator or modifier of carcinogenesis. 


of their efforts have inevitably made their way into regulatory decisionmaking. 
These short-term methods include a variety of in vitro (i.e., outside of a living 
organism) tests in various stages of development or use. They typically measure 
chemical properties that are thought to be related to the likely mode of action 
of carcinogens. The most common tests measure either genetic mutation in 
bacteria^^'' or so-called "transformation" of mammalian cells maintained in 
tissue culture to a state in which their growth behavior and surface properties 
resemble those of cancer cells. ^^"^ Other short-term tests for carcinogenicity are 
under development.^'*'' 

The best known of the short-term tests is the Ames test for bacterial mu- 
tagenicity."''^ The test uses a specially constructed strain of bacteria which or- 
dinarily will not grow without certain special nutrients. These bacteria, after 
treatment with the test chemical, are placed in a mixture lacking the special 
nutrients. Since only mutant bacteria which have acquired new functional ability 
can grow in the deficient media, the bacteria growth measures the ability of the 
test chemical to mutate the bacterial DNA. While there is a high correlation 
between a chemical's ability to mutate bacterial DNA in this test and its ability 
to produce tumors in animals, -''^ the correlation is not perfect; roughly ten percent 
of the chemicals which Ames found mutagenic did not prove positive in bioas- 
says.^^^ Positive results in the Ames test or in one or more other short-term tests 
have not generally been regarded by regulators as decisive predictors of human 
hazard, but they often have been cited as confirmatory of other evidence of 
carcinogenicity. ^^'^ 

Because positive findings in short-term tests have been shown to correlate 
well with evidence from animal experiments, and because the tests are rapid and 
inexpensive, they have been relied on increasingly by industry in determining 

253. See R. Deverat, Bacterial Tests for Potential Carcinogens 241 Scientific American 40 
(1979); Venett, Bacterial Mutation as an Indication of Carcinogenicity, 36 But. Med. Bull. 57-62 

254. See R. J. Pienta. J. A. Poiley. and W. B. Lobherz III, Morphological Transformation 
of Early Passage Golden Syrian Hamster Embryo Cells Derived from Cryopuserved Primary Cultures 
as a Reliable In Vitre Bioassay for Identifying Diverse Carcinogens, 19 Intemat'l J. of Cancer 642- 
655 (1977). 

255. See T. Maugh, Chemical Carcinogens: The Scientific Basis for Regulation, 201 Science 
1200, 1203-4 (1978). See OSHA Carcinogen Policy 5162-5164; Lutz, Invivo Covalent Binding of 
Organic Chemicals to DNA as a Quantitature Indicator in the Process of Chemical Carcinogenesis, 
65 Mut. Res. 289 (1979). 

256. B. Ames, Identifying Environmental Chemical Carcinogens, 204 Science 587, 1979. 

257. In a validity study of 300 chemicals, roughly 90 percent of the identified animal car- 
cinogens gave positive results in the Ames test. McCann, J. E., Chois, E., Yamosaki and B. Ames, 
Detection of Carcinogens as Mutagens Notice Salmonellal Microsome Test: Assays of 300 Chemicals, 
72 Proc. National Acad, of Sciences 5135-5139 (1975); J. McCann and B. Ames. 73 Proc. Nat. 
Acad. Science 950 (1976). In addition to confirming the relevance of Ames tests results, this 
observation is powerful evidence that genetic alterations are an important event in carcinogenesis. 

258. In other words, although non-carcinogenic the chemicals nevertheless mutate bacterial 
DNA. The incidence of * 'false negatives." i.e.. those results in which chemicals that cause cancer 
in animal tests do not mutate bacterial DNA was also ten percent. McCann. et. al, supra note 289. 

259. OSHA Carcinogen Policy at 5173. 


which chemicals to develop and by regulatory agencies in determining which 
chemicals should be tested further. ^^^ Short-term tests can be performed in a 
few days or at most a few months, ^^' and individually rarely cost more than 
$1,000.^^^ A battery of different short-term tests can facilitate rapid screening 
of the large number of chemicals to which humans are exposed. ^^^ Short-term 
tests can also aid in ranking chemical carcinogens according to their relative 
potency. None of the current tests, however, yields data for quantitative pre- 
dictions of human risk. 

5. Chemical Structure. 

Prior knowledge of the physical and chemical properties of a substance can 
provide limited information about its capacity to cause cancer. Scientific literature 
on the association between carcinogenicity and the structural and functional 
characteristics of chemicals is extensive. ^^"^ For example, chemicals capable of 
causing mutations are thought to belong to a group called electrophiles; deter- 
mining that a chemical falls within that group supports the suspicion that it may 
be mutagenic and, thus, potentially carcinogenic. ^^^ Like the short-term tests, 
knowledge of chemical and physical properties can be used in screening chem- 
icals and thus aid in setting priorities for further testing. ^^^ But such knowledge 
alone cannot support predictions of carcinogenicity and it therefore has never 
been key to a decision to control exposure. 

6. Determining Human Exposure. 

a. General approach. Even a chemical capable of causing cancer cannot 
harm humans if humans are not exposed to it. The risk posed by a carcinogen 
is a function of the number of people exposed to it and the route and magnitude 
of their exposure, as well as of the potency of the substance itself. Accordingly, 
some inquiry into exposure is an integral part of regulatory decision making. 
Techniques for measurng or predicting exposure to chemicals in the environment 
are, however, more varied and generally less systematic than the experimental 
methods for evaluating carcinogenicity. ^^^ Perhaps for this reason, exposure 
issues have often been treated more casually in agency decisions. But where 
Congress has instructed regulators to assess the costs of regulation as well as 
the health benefits, they have had to devote increasing attention to such issues. 

260. The Ames test, for example, can provide a quantitative measure of the potency of a 
mutagen. R. Devoret, Bacterial Tests for Potential Carcinogens, 241 Scientific American 40, 45 

261. OTA Report at 115. 

262. Id. 

263. OTA Report at 121-22. 

264. See generally, L. Slesen and R. Sardler, Categorization of Chemicals Under the Toxic 
Substances Control Act, 1 Ecology L.Q. 359, 374-82 (1978). 

265. Id. at 362-63; OTA Report at 1 16-20. 

266. OSHA Carcinogen Policy at 5178. OSHA said it would use evidence of "structural 
similarity" in priority setting. See OTA Report at 21. 

267. See OTA Report at 145; NAS, Risk Assessment in the Federal Government: Managing 
the Process (1983). 


In the regulatory context, the term "exposure" has multiple meanings. 
Often it is used to describe a setting in which humans enounter a substance. 
Thus an exposure to a carcinogen can be a particular context (at work) or medium 
(in food) in which humans come into contact with the substance. Whether an 
exposure to a carcinogen involves a risk of cancer and, if so, what the magnitude 
of that risk is are functions of other characteristics of the exposure, specifically 
the route by which humans encounter the substance and the quantity they are 
likely to absorb. 

Because the data suggesting that a substance can cause cancer usually are 
the product of animal feeding studies, it could be inferred that the substance 
may have to be ingested to cause cancer. ^^^ One might question the relevance 
of such evidence in evaluating the risk to humans posed by other routes of 
exposure. It is clear, however, that substances that are inhaled or applied topically 
can be absorbed and metabolized. At the same time, even when the route of 
human exposure duplicates the laboratory model, there may be differences in 
the way animals and humans rnetabolize the substance that could affect assess- 
ment of the risk to man. 

The regulator's inquiry should be to determine whether humans encounter 
a carcinogenic material by a route that permits it to enter their systems and reach 
organs that are vulnerable to its effects. But verification of human uptake alone 
may not justify regulation. The next level of analysis, theoretically, is to deter- 
mine the effective dose to which humans are exposed. This may involve a dual 
inquiry: a determination of the levels of the chemical present in the environment, 
e.g., in the workplace or in food, and a determination of the levels of human 
uptake. Agencies routinely must make substantial compromises in their efforts 
to answer these questions. 

While the evaluation of human exposure to environmental carcinogens can 
assume broad scope and sophistication, in some regulatory programs it requires 
no effort at all. If a chemical is submitted for approval as a food additive, for 
example, FDA can confidently assume that some humans will ingest it. Since 
the Delaney Clause forbids approval of any carcinogen as a food additive, no 
further inquiry is necessary to justify regulatory action. Sometimes the exposure 
issue arises as a question of agency jurisdiction. For example, FDA has re- 
sponsibility for regulating indirect food additives — substances used in contact 
with food — but if such a substance is not "reasonably expected to become a 
component" of food, it is not subject to the agency's jurisdiction. -^^^ The issues 
whether a packaging material migrates into food thus has legal as well as health 

Each exposure to a carcinogen can in theory be characterized to reflect (1) 
the levels, frequency, and duration of exposure; (2) the form of the substance; 
(3) the route of human exposure; and (4) the likelihood of human uptake of the 
material. Careful assessment of the health risk posed by exposure to a carcinogen 
would require exploration of each element. For example, to assess accurately 

268. OTA Report at 126. 

269. 21 U.S.C. § 321(s); Merrill, Carcinogens in Food at 222-26. 


the risk posed by asbestos linings in hand-held hairdryers, the CPSC would have 
to determine, first, whether asbestos was released during use; second, how much 
asbestos was released each time a dryer was used and how often, on average, 
dryers were used; third, the form of any released asbestos; and fourth, the 
likelihood that released particles would be inhaled by users of hairdryers or 
members of their households. Verifying each feature of this one type of exposure 
to asbestos could obviously be a protracted undertaking. ^^^ 

An agency's task is simplified if the levels of a chemical accessible to 
humans are known, e.g., as is the case for food additives, and the route of 
exposure associated with cancer induction parallels the route of exposure to be 
regulated. Sometimes levels of a chemical in the environment are easily estab- 
lished but the route of human exposure differs from routes evaluated experi- 
mentally. Then the agency must determine or estimate levels that are taken up 
and compare them with the experimental doses. When a chemical is present in 
the ambient environment, an agency usually attempts to measure concentrations 
at various locations over time to determine average exposures and depletion 
patterns. With workplace exposures, it may also be possible to verify actual 
absorption through physical examination of exposed employees. But in the case 
of episodic and possibly widely varying exposures, e.g., from asbestos in hand- 
held hairdryers, estimation of ambient levels may be possible only through 
simulation and determining acutal absorption may be virtually impossible. -^^^ 

b. Agency performance. Data describing the wide variety of human ex- 
posures to chemicals is of uneven quality. Current limits of analytical chemistry 
preclude routine monitoring of some exposures. Some sophisticated computer 
models have been developed for predicting, from emissions data and information 
about meterological and terrestrial conditions, the magnitude and duration of 
human exposure to chemicals distributed in air and water. But some substances 
are encountered in distinctive ways, making measurement of particular human 
exposures a unique and sometimes costly undertaking. FDA has access to food 
consumption surveys conducted by the U.S. Department of Agriculture and has 
commissioned studies of dietary habits. These data permit rough estimates of 
exposure to intentional ingredients of common foods; they are less useful in 
predicting atypical consumption patterns. Furthermore, they have few counter- 
parts. Other agencies typically rely on a combination of environmental moni- 
toring, laboratory testing, and occasionally sheer hypothesis. 

The care with which agencies assess chemical exposures depends on several 
circumstances. An agency is more likely to go beyond a back-of-the-envelope 
estimate if methods for measuring exposure have been developed or monitoring 
data are available. It is, obviously, more likely to perform a careful assessment 
if the governing legal standard requires refinement or it is able to require man- 
ufacturers or distributors of a substance to generate necessary data. In most of 
the programs studied here, however, the agency bears the burden of gathering 

270. Generally, empirical verification of exposures requires new investigations, for rarely is 
there any independent record of consumption or use patterns. 

271. See Merrill, Consumer Products at 1272-73, 1346-51. 


exposure data. Agency exposure estimates, accordingly, have often been per- 

The four agencies studied have rarely disagreed in their assessment of a 
substance's carcinogenicity but they have often appeared to diverge in their 
assessments of the magnitude of the risks posed by a chemical. ^^^ Such differ- 
ences have usually reflected differences in the exposures for which are responsible 
and in the legal standards they must apply. No actual episode provides the ideal 
example, but a hypothetical case can illustrate this phenomenon. Suppose that 
a substance used as an ingredient in food, as a component of consumer products, 
and as a pesticide is found to be carcinogenic. The four agencies might respond 
to this finding in different ways. EPA, for example, might conclude that neither 
the exposures experienced by pesticide applicators nor consumers of food posed 
a serious risk, and decline to restrict usage. FDA, by contrast, would ban the 
substance as a food additive. ^^^ The CPSC could discover that the substance 
was incorporated in consumer products in such a way that human uptake was 
unlikely and thus conclude that any risk was negligible. ^^"^ OSHA, meanwhile, 
would presumably require that the levels to which production workers were 
exposed be reduced to the lowest feasible level. -^^^ 

It should not be surprising that the four agencies might assess the health 
risk of the same chemical differently. In reality, the agencies regulate uses and 
exposures, not chemicals; their jurisdiction is defined by chemical uses or ex- 
posure contexts. It would be surprising if a chemical encountered in several 
different contexts posed a comparable risk in each one. Exposures are likely to 
vary in route, level, duration, and frequency; the form of the chemical may vary 
as well. These differences may warrant radically different assessments of risk, 
wholly apart from differences in outcome dictated by statutory criteria. 

The four agencies also differ in their approach to assessing exposure. The 
discovery of a food-packaging material in food may conclude FDA's inquiry 
into the issue of human exposure. ^^^ In regulating environmental contaminants, 
however, the agency has also attempted to measure levels of contamination and 
estimate frequency of consumption, primary because it interprets the statute as 
requiring some balancing of costs of control. ^^^ The agency generally performs 
its own tests to determine contamination levels, but it has relied primarily on 
existing food consumption data to estimate average individual exposure. For 
FDA, route of exposure has not often been an issue. "^^^ 

In many proceedings, the CPSC has rested on the discovery that a substance 
present in consumer products is accessible to consumers during ordinary use.^^^ 

272. See "Low Growth," Wall Street Journal, October 10, 1976 (discussing the varied 
responses of the several agencies to acrylonitrite). 

273. See Merrill, Regulating Carcinogens in Food at 178-84. 

274. 43 Fed. Reg. 25661, 25664, (1978); Merrill, Consumer Products at 1298. 

275. See OSHA Draft at 77, 79. 

276. This need not be so. See Monsanto Corp. v. Kennedy, 613 F.2d 947 (D.C. Cir. 1979). 

277. Merrill and Schewel, Environmental Contaminants at 1379-82. 

278. But see 45 Fed. Reg. 721 12 (1980) (lead acetate in hair dyes). 

279. See Merrill, Regulating Consumer Products at 1349. 


Route of exposure has not been a critical issue. In some recent cases, however, 
the Commission has not been content to rely merely on the fact of exposure and 
has commissioned studies to determine extent and duration. These studies have 
usually involved simulated use of a product in laboratory settings, where con- 
centrations can be reliably measured over time. 

Of the four agencies, OSHA appears best positioned to estimate actual 
levels of human exposure to toxic materials. Not only can its inspectors monitor 
industrial establishments, but it can call on investigations by NIOSH and, in- 
creasingly, on exposure records kept by employers. It has, however, often been 
impeded by the difficulty of determining exposure levels from years past, a 
persistent obstacle to the conduct of epidemiological studies. ^^^ 

EPA may have the most difficuh task of assessing exposures because humans 
encounter pesticides in so many different contexts. Consumers may ingest small 
amounts on target crops as well as on other commodities contaminated by drift 
during application or through environmental transfer. Residents of localities 
where pesticides or herbicides are used frequently may encounter ambient levels 
high enough to be of concern. Pesticide applicators encounter the highest levels 
but may also be better equipped to control exposure through cautious use and 
by protective clothing. EPA thus must often consider several different populations 
whose exposures vary in every dimension, and assess the risk for each before 
deciding whether a pesticide causes "unreasonable adverse effects. "^^' 

c. Analytic methods. An agency's ability to measure human exposure to 
a carcinogen, whether for the purpose of assessing the health risk or monitoring 
compliance with regulations, is limited by the capability of chemical analysis. 
This technology has witnessed genuinely extraordinary advances during the last 
two decades. Sophisticated analytic procedures, including gas and liquid chro- 
matography and radio-labeled isotopes, have permitted analysts to detect and 
measure compounds at extremely low levels. Indeed, this improvement in an- 
alytic technology that has significantly enlarged the universe of potential car- 
cinogens to which humans are exposed — by revealing their presence in materials 
or environments for the first time. 

Despite remarkable advances, however, the technology still has limits. Some 
chemicals cannot be distinguished at extremely low levels in many media. Meth- 
ods may exist for verifying a chemical's presence, but concentrations may be 
below the limit of measurement, impeding quantitative assessment. Occasionally, 
the agent responsible for cancer may not be known. For example, the constituent 
of coke oven emissions that produces an increased incidence of lung cancer 
among exposed steelworkers has not been identified. OSHA therefore selected 
another easily measured constituent of coke-oven emissions as a marker for 
monitoring compliance. ^^^ Accordingly, an agency must match its regulatory 
requirements to materials that can be detected and specify control levels that are 
within the range of reliable measurement. 

280. See discussion of the study of coke oven emissions at text accompanying notes 210- 

281. See NAS, Regulating Pesticides 66-89 (1980). 

282. See 41 Fed. Reg. 46742 (1976). 


Just as the capability of analytic methods can restrict an agency's ability to 
measure exposure to carcinogens, improvements in chemical analysis can en- 
hance its capacity to evaluate and control exposure. Indeed, such improvements 
may reveal health hazards not previously recognized. For example, progressive 
refinements in analytic methods eventually revealed residues of DES in marketed 
cattle to which the drug had been administered. ^^-^ A similar phenomenon has 
forced FDA to reassess its approach to regulating packaging materials that migrate 
to food.^^"^ Here analytical improvements have enlarged FDA's regulatory ju- 
risdiction by confirming previously unsuspected or ignored residues. 

E. Quantitative Assessment of Risk 

1 . Background. 

Some of the most intensely debated issues in health regulation surround the 
efforts to develop quantitative estimates of the risk posed by specific substances. 
A variety of incentives have encouraged regulators to devise methods for quan- 
tifying risk. Quantification permits differentiation among chemicals that are 
potential candidates for regulation and, thus, aids in setting priorities. Numerical 
depiction of the risks of particular chemicals can be used to illustrate the health 
benefits obtainable from controlling human exposures. Perhaps most important, 
and certainly most controversial, quantitative risk assessment facilitates analysis 
of the costs and benefits of alternative control strategies. 

Many observers have criticized quantification, particularly if used to set 
exposure limits for carcinogens. They call attention to the wide disparities in 
the risks predicted by different models. They stress, as well, the lack of empirical 
verification of any of the models and the uncertainties inherent in extrapolating 
from animals to humans. ^^^ Supporters of quantitative risk assessment acknowl- 
edge these deficiencies, but defend its use as a means of structuring decisions 
that inevitably involve substantial uncertainties. They further contend that in- 
telligent evaluation of efforts to reduce human exposure to carcinogens requires 
some method for estimating the magnitude of the health consequences of alter- 
native control measures. ^^^ 

Broadly defined, quantitative risk assessment is a method or process of 
calculating the estimated likelihood that a particular exposure to a substance will 
cause cancer. While the mathematical models available to perform such a cal- 
culation vary in both their operation and outcomes, they start from the common 
assumption that the likelihood of harm from exposure to a toxic substance is a 
function of two variables: the potency of the substance and the dose, the latter 
characterized by the level, route, and duration of exposure. In this respect, the 
assumptions underlying quantitative risk assessment for carcinogens do not in 
principle differ from those historically relied on in evaluating other types of toxic 

283. See Merrill and Hutt at 485-96. 

284. See Monsanto Corp. v. Kennedy, supra note 276; Merrill and Hutt at 86-95. 

285. See OSHA Carcinogen Policy at 3189-95. 

286. See Lave, The Strategy of Social Regulation 31-34 (1981). 


Government agencies have used quantitative techniques in regulating haz- 
ardous materials for several decades. Like the contemporary extrapolation models 
for carcinogens, these techniques assumed that the capacity of a substance to 
harm an organism was a function of the dose; lower doses were likely to be 
safer than higher ones. Traditional regulation of non-carcinogens relied on so- 
called "safety factors," which were applied in the following way. On the premise 
that many cells had to be damaged before an organism sustained permanent 
harm, toxicologists conducted tests to determine the highest dosage that produced 
no adverse effects in experimental animals. ^^^ The regulator would divide this 
"no observed effect" level by a predetermined number, i.e., a safety factor, to 
arrive at a dose predicted to be without risk for humans. For direct food additives, 
for example, FDA ordinarily used a safety factor of 100.^^^ While there were 
attempts to justify these safety factors biologically, most toxicologists would 
concede that the concept was simply an attempt to account for the possibility 
that humans might be more sensitive than the species tested, and that the precise 
numbers used were selected for reasons of convenience. 

It should be stressed that this approach to quantification was not designed 
to estimate the magnitude of the risk associated with exposure to a chemical. 
Rather, the approach rested on the assumption that toxic materials had thresholds. 
The "no observed effect level" was an approximation of this threshold for the 
species tested; the safety factor accounted for the possibility that humans, being 
more heterogeneous, might also be more vulnerable. The permitted exposure 
level derived by this method was accepted, probably literally as well as opera- 
tionally, as "safe." Although regulatory agencies occasionally used safety fac- 
tors in regulating carcinogens, ^^^ it was recognized that this approach was 
inconsistent with the no threshold hypothesis that dominates current thinking. 

Other quantitative approaches are illustrated by FDA's and EPA's evolving 
efforts to regulate residues of carcinogens used in the marketing or production 
of food. Animal drugs, indirect food additives, and pesticides present analogous 
regulatory problems. It was never the objective of producers that these substances 
should become part of the food people consumed, yet it has been recognized 
that some carryover was likely. The objective of regulation has been to limit 
any residues to "safe" levels. For carcinogens, the Delaney Clause theoretically 
would forbid any residues at all. But this policy has always been subject to the 
practical constraint that FDA cannot prohibit what it cannot detect. 

In choosing methods for detecting animal drug residues, for example, FDA 
has long been confronted with the reality that any choice effectively sanctions 
residues the method chosen cannot find. For several years, the agency set a 
numerical figure, of 2 parts per billion, as a numerical target for regulatory 

287. Kimm et al., "Waterbome Carcinogens: A Regulator's View," in Crandall and Lave, 
eds., The Scientific Basis of Health and Safety Regulation, at 229 (1981). 

288. Merrill, Carcinogens in Food at 207. 

289. FDA sometimes used a safety factor of 1000 in determining negligible migration levels 
for indirect food additives that might cause cancer. See correspondence to the author from Joseph 
V. Rodricks, then Assistant to the Director for Science Policy, FDA Bureau of Foods, September 
28, 1978. 


assays for carcinogens. Like the safety factor approach, FDA's selection of 2 
ppb did not represent an effort to quantify the risk associated with exposure to 
low levels of carcinogens; rather, it was an expression of the agency's willingness 
to ignore lower exposures. While ostensibly justified on toxicological grounds, 
this approach was dictated by necessity and convenience. FDA never attempted 
publicly to reconcile it with the no threshold hypothesis, to which the agency 
had consistently adhered. 

Unlike these numerical approaches for setting exposure limits for toxic 
materials, the methods for quantitative risk estimation currently being used by, 
or urged upon, regulatory agencies are designed to illuminate the probable con- 
sequences of alternative control measures and, in some cases, to identify risks 
that can be considered insignificant. Subsequent sections describe the methods, 
summarize their scientific underpinnings, and analyze their strengths and weak- 

2. Estimating Cancer Risks.. 

Assessment of the magnitude of the human cancer risk presented by a 
carcinogen requires two types of information and a method for relating them. 
The first is information about the incidence of cancer produced by different doses 
of substance, i.e. , a dose-response curve. The second type of information needed 
is information about the exposures humans are encountering or are likely to 
encounter. There are perhaps half a dozen mathematical extrapolation models 
that can be used to relate these two types of information. Some such model is 
necessary because the information about potency and exposure are almost in- 
variably derived from different sources. 

Human epidemiological studies often, perhaps usually, are not sensitive 
enough to detect the effects of prevailing levels of exposure to most substances 
that cause cancer. ^'^^ In the majority of cases, therefore, agencies must make 
decisions on the basis of data derived from experiments with laboratory animals. 
But standard animal bioassays, for both economic and scientific reasons previ- 
ously outlined, ^^' employ doses that proportionally are much higher than those 
to which humans are exposed. Some method is needed to translate these data 
into realistic estimates of the magnitude of the risk to humans. 

Accordingly, quantitative risk assessment usually involves two extrapola- 
tions, both of which introduce uncertainties. The first extrapolation is from 
laboratory animals to humans; the second is from the high doses observed to 
cause cancer in animals to likely human exposure levels below the range for 
which data are available. Actually, where both extrapolations are being per- 
form.ed, there is a third source of uncertainty. For we are asked to assume not 
only that magnitude of risk is a function of dose and that animals are reliable 
qualitative predictors of potential human hazard, but also that the dose-response 
experience of animals is likely to replicate that of man. The premise of the first 

290. Hoover and Strasser, Artificial Sweeteners and Human Bladder Cancer, 1 Lancet 837 

291. See text accompanying notes 236-40 supra. 


extrapolation — that an agent which induces cancer in other mammals presents 
a potential risk for man — is well-accepted. ^*^^ There is also considerable evidence 
that, within species, cancer incidence generally increases with dose. Physiolog- 
ical differences between experimental animals and man, however, weaken con- 
fidence in quantitative extrapolation between the species. ^^-^ 

If a regulator desires to quantify the risk associated with human exposure 
to animal carcinogens, however, this extrapolation is unavoidable. To perform 
this extrapolation, some adjustment must be made for the obvious disparity in 
the sizes of the two species to posit biologically comparable human doses. This 
adjustment is reflected in a so-called "scahng factor" by which animal dose- 
response data are translated into ostensibly equivalent human doses. Several 
different approaches have been advanced as biologically rational. ^^"^ The scaling 
factors currently popular with regulatory agencies make comparisons on the basis 
of dose per unit of surface area, e.g., micrograms per square centimeter, ^^^ or 
on the basis of relative dietary concentrations. ^^^ A third scaling factor, originally 
quite popular, bases comparisons on the relative body weights of the two spe- 
cies. ^^^ The current preference for the other approaches reflects official conser- 
vatism; the use of relative surface area has the effect of increasing estimates of 
human risk because the disparity in surface area between rodents and humans 
is relatively less than the disparity in their weights. ^^^ What little empirical 
evidence exists, however, suggests that all four of the common scaling factors 
may modestly overstate human risk. 

The attempt to estimate the effects of low doses of a carcinogen from its 
observed effects at high doses is more controversial than the qualitative extrap- 
olation from animals to man.^^^ To make this extrapolation with confidence an 
agency would need, but almost never has, knowledge of the underlying mech- 
anism by which the substance causes cancer. Moreover, in most cases there are 
no data from animal exposures at doses corresponding to the human exposure 
levels that are of regulatory interest. Thus an agency must have some way of 
estimating risks at doses for which no experimental data exist. In the absence 
of data, quantitative risk assessment requires that an agency select some bio- 
statistical model for estimating low dose effects. Several mathematical models 
have been formulated for this purpose. ^^° While the models vary widely in their 
biological rationales, ^^' all of them purport to describe the relationship between 

292. OTA Report at 122; OSHA Carcinogen Policy at 5190-95. 

293. OTA Report at 170. 

294. Id. at 169-70; Letter to author from Joseph V. Rodricks, Clement Associates, July 9, 

295. See OSHA Carcinogen Policy at 5 189; P. Hutt. Unresolved Issues in the Conflict Between 
Individual Freedom and Government Control of Food Safety, 33 FDC L.J. 558, 578 (1978). 

296. Rodricks Letter, supra note 336. 

297. Id.; OTA Report at 169. 

298. Letter to the author from Marvin Schneiderman, Ph.D., March 28, 1981. 

299. See OSHA Carcinogen Policy at 5179; See generally, OTA Report at 157-72. 

300. OTA Report at 160-62. 

301. Rodricks Letter, 5Mprfl note 294. 


recorded doses of a carcinogen and the observed incidence of tumors, and then 
to predict the probable relationship at lower doses. 

Most extrapolation models make assumptions about the stochastic properties 
of the interactions between carcinogens and target organisms. One procedure 
relies on the classic dose-response relationship derived from pharmacologic stud- 
ies, which is an S shaped curve, indicating that some organisms in an exposed 
population are extremely sensitive to toxic agents and others are highly resis- 
tent.^^^ This relationship led Mantel and Bryan to hypothesize that the dose- 
response function of a carcinogen can be approximated by a probit curve in the 
low dose range. ^^^ Another model starts from the premise that no effect can 
occur when the dose of a carcinogen is zero and estimates the effects of low 
exposures by a linear extrapolation from observed positive doses to zero. A third, 
the so-called "one hit" model, assumes that a single "hit" of a carcinogen can 
produce a positive response. Any molecule of the substance is regarded as having 
the same probability of causing the reaction; the cumulative probability of cancer 
is thus a function of cumulative exposure. ^^"^ The linear model predicts direct 
straight-line proportionality between dose and effect until the dose decreases to 
zero.^^^ The one-hit model generally produces higher estimates of risk at low 
doses. ^^^ None of the models that have gained popularity, however, suggests a 
threshold dose below which a carcinogen poses no risk of cancer. 

Because several extrapolation models enjoy respectable support, a regulator 
is faced with a difficult choice at the outset. One approach would be to select 
an extrapolation model on the basis of correspondence with the data in the 
observable, i.e., high-dose, range. If one model fit the observable data better 
than others, it might be favored as the best predictor of effects at low doses. 
Choosing among extrapolation models based on the closeness of "fit" to the 
•available data is rarely possible, however, because the data usually fit several 
models equally well.^^^ The inadequacy of closeness of fit as a criterion for 
selecting an extrapolation model is illustrated by the results of applying several 
models to the data derived from a unique animal experiment performed by the 
National Center for Toxicological Research. ^^^ This so-called "megamouse study" 
used 24,000 rodents to measure the effects of 2-acetyl-animofluorine (2-AAF), 
a known carcinogen, administered at doses an order of magnitude lower than 
those customarily used in animal bioassays. The final data from the study were 
statistically fit to the five best-known mathematical models to determine whether 
"closeness of fit" would provide a basis for preference. While two models could 
be eliminated, each of the other three was equally compatible with the observed 

302. See Mantel and Bryan, Safety Testing of Carcinogenic Agents, 27 J. Nat'l Cancer Inst. 
(August 1980). 

303. Id. 

304. OTA Report at 161-62. 

305. Id. 

306. The probit model does not predict linear risk at low doses. 

307. See OSHA Carcinogen Policy at 5184-85; OTA Report at 162. 

308. See Bruce, Low Dose Extrapolation and Risk Assessment (1980). 

309. Id. at 6. 


The choice of an extrapolation model can significantly affect risk estimates. 
While the familiar models often fail to reveal differences at high doses, they 
often predict sharply different risks in the low-dose range. This disagreement 
among the models is illustrated numerically and graphically in Tables 2 and 3, 
which are based on the NCTR study. ^'^ 

Because it is usually not feasible to prefer one extrapolation model based 
on its compatibility with observed data, regulatory agencies have relied on other 
criteria. For a time some espoused the linear model, which assumes that the 
incidence of cancer will be directly proportional to dose at low exposures.^^' 

Table II. 

Probability of Observing No Tumors in an Experiment Involving A^ 
Animals with a Food Chemical Affecting 1% of Population 

A^ Probability (%) 

500 1 

250 8 

100 37 

50 61 

25 78 

10 90 

1 99 

This table illustrates the weakness of animal bioassays in establishing a finding of no effect. 
When testing a chemical assumed to cause tumors in 1% of the population, a routine group size of 
50 would have a 37% chance of showing no tumors. Thus in a single test a "negative result" of 
no observed tumors would be compatible with both no effect at all or in 61% of the tests with a 
1% incidence of tumors. With a group size of 500 the chances of "missing" a carcinogen of the 
same potency drops to only 1%. making the negative finding more meaningful, but 500 is an 
outlandish group size for routine animal bioassays. 

Table III 
Range of Estimate "Virtually Safe" Doses* 



Safe Doses 

. ppb 



















*"Virtually safe" defined as a risk of tumor no greater than 1 in 1 million over a lifetime (1/10- 
6), extrapolated from a study in which mice were administered 2-AAF in the diet for 24 months. 
Estimates are based on 99 per cent confidence level. 

310. Id. (Table C is taken from Table 2 in Bruce, Table D from Table 7a). 

311. E.g. , IRLG Scientific Principles at 259. 


One reason for preferring this model is its so-called "conservatism," i.e., the 
model is less likely than others to underestimate the numbers of cancers expected 
as the result of exposure to low doses of a carcinogen.^'^ If this were the sole 
concern, an agency would always select the model that produced the highest 
estimated risk of cancer so that regulation could prevent the hypothesized "worst 
case." Regulators, however, have generally tended to seek one extrapolation 
procedure that can be followed in evaluating all carcinogens. This is not essential; 
it has been suggested that regulators should use several different models to display 
the range of potential risk associated with exposure to a carcinogen. In the absence 
of a powerful desire for flexibility, however, consistency and predictability argue 
for the selection of a single model so that risk estimates for different substances 
are readily comparable. This approach facilitates ranking of carcinogens. The 
agency with the most experience in quantitative risk estimation for carcinogens, 
EPA, currently favors the multi-stage model, which is claimed to best represent 
the mechanism of carcinogenesis.^'^ Other agencies are likely to follow EPA's 

Another way of estimating the effects of low doses is to relate the length 
of time between exposure and tumor occurrence to experimental dose.^'"^ This 
correlation is referred as "time to tumor." The assumption underlying this 
approach is that not only are very low doses likely to produce few cancers, but 
most would not be expected to occur during the normal life span of exposed 
individuals. Routine bioassays do not generally yield time-to-tumor data, how- 
ever, and agencies have rarely relied on this refinement in quantitative risk 
estimates. Moreover, recent scholarship suggests that the results of this approach 
will not differ significantly from those of purely binary models. "^'^ 

Empirical support for quantitative extrapolation is concededly relatively 
sparse, but not negligible. For most human carcinogens there is corroborative 
animal data,^'^ and for a few, e.g., ionizing radiation,'''^ a linear no-threshold 
curve seems to fit observed human experience. However, benzene and arsenic, 
both human carcinogens, apparently contradict both necessary extrapolations,^'^ 
suggesting that neither can be relied on with complete confidence. There have 
been few systematic efforts to compare estimates of cancer incidence derived 

312. Id. 

313. GMA Petition, supra at 120. 

314. OTA Report at 164. 

315. See K. Crump, Models for Carcinogenic Risk Assessment, 202 Science 1 106 (1978). 

316. OTA Report at 192. 

317. See Lave, The Strategy for Social Regulation 1 13-1 14 (1981). Analogies drawn between 
physical agents and chemical carcinogens should be cautiously regarded. The entry of radiation into 
cells is governed by physically predictable laws while chemicals must undergo a series of poorly 
understood transport and metabolism steps before the critical interaction with the target cells. See, 
Maugh, Chemical Carcinogens: How Dangerous are Low Doses, 202 Science 37, 37 (1978). 

318. See note 226 supra; OTA Report at 141-42. There is a preliminary report that animals 
exposed to high doses of benzene develop leukemia, but only in conjunction with a toxic noncar- 
cinogenic effect, aplastic anemia. This result suggests that the leukemia is a secondary effect of the 
aplasia, and that the threshold for the aplasia would also be the threshold for the leukemia. See R. 
Wilson, Comparison of Carcinogenic Potency in Animals and Man, (1978); OTA Reports at 170- 


from animal data with actual human experience, mainly because data are so 
fragmentary. One study did compare predicted and actual human cancer rates 
for six substances known to be carcinogenic for both animals and man.^'^ The 
authors found a close correspondence for three of the substances and a fair 
correspondence for a founh, aflatoxin. For the remaining two, the animal data 
substantially overestimated human risk. 

3. Depiction of Cancer Risks. 

Once an agency has selected an extrapolation model, it can be applied to 
the available toxicological data to derive estimates of the human cancer risk at 
different exposure levels. At this stage, it is not essential that the agency have 
information about actual human exposures; an assessment can be expressed in 
terms of risk per unit dose, as Table 3 illustrates. If the agency desires to estimate 
the cancer risk associated with a particular type of exposure, e.g., the risk of 
consuming saccharin-sweetened beverages, it must develop information about 
the amount of saccharin added to each drink, average daily consumption of 
saccharin-sweetened beverages, and duration of consumption. Often an agency 
must make assumptions about probable exposure patterns because reliable data 
are unavailable. Based on such assumptions, or on available data, the agency 
can derive an estimate of the human cancer risk posed by an animal carcinogen. 

Quantitative risk estimates of this kind are usually presented in one of two 
forms: as estimates of average individual risk or as estimates of possible cancer 
incidence in the exposed population. When FDA proposed to ban saccharin, for 
example, it estimated, based on the now-famous Canadian study in rats and 
using the linear model, that consumption of artifically sweetened soft drinks 
increases one's individual risk of bladder cancer by 0.4 percent (from 1.50 to 
1 .54 percent). The agency also estimated that the additional number of saccharin- 
caused bladder cancers in the consuming population could range as high as 

One feature of these estimates should be carefully noted. Each represented 
a "worst case" depiction of the risk of consuming saccharin. From a range of 
estimates statistically supportable by the underlying data, FDA reported the 
highest figure. In addition, the agency's assessments consciously incorporated 
exaggerated assumptions about the number of persons who consumed saccharin- 
sweetened beverages and the amounts they consumed. The resulting risk esti- 
mates were termed conservative, in that by design they conveyed the maximum 
risk of saccharin consumption. 

The practice of using the upper confidence limit of extrapolation results to 
express estimates of human risk is not unique to FDA, but it has been criticized 
by industry groups as well as by more objective observers of agency decision- 

319. See R. Wilson, supra note 319; see also NAS/NRC Environmental Studies Board, 
Carcinogenesis in Man and Laboratory Animals, in Pest Control: An Assessment of Present and 
Alternative Technologies, Vol. I of Contemporary Pest Control Practices and Prospects pp. 66, 82 

320. Saccharin and Its Salts, 42 Fed. Reg. 19995 (1977). 


making.''^' These critics argue that the practice distorts quantitative estimates of 
risk. This distortion is further exaggerated if the agency also builds in other 
conservative assumptions, which are multiplied as analysis proceeds. "^^^ One 
response to these criticisms is that quantitative risk assessment inevitably involves 
so many fragile assumptions and uncertainties that prudence dictates a posture 
favoring human health. 

It should be emphasized that the numbers produced by quantitative risk 
assessment are merely estimates of the potential effects of human exposure to a 
carcinogen; they are not predictions of the number of cancers that the substance 
will cause. First, the estimates assume a correlation between potency in animals 
and potency in man that is not verified for most substances. Second, the toxi- 
cological data underlying such estimates is often sparse — occasionally only a 
single study in one species — and thus may not be representative of other species. 
Finally, the agency has had to extrapolate from data at doses far above those of 
immediate concern, using models whose biological soundness is unknown. 

4. Uses of Quantitative Risk Assessment. 

Despite its limitations, quantitative risk estimation has exerted a strong 
appeal on both analysts and regulators. Without some means of describing the 
magnitude of the health effects associated with exposure to a carcinogen — or of 
quantifying the health benefits of attempts to reduce exposure — an agency must 
find some other basis for deciding what controls to require. Legislation occa- 
sionally provides an answer. The Delaney Clause, for example, makes quanti- 
fication of the risk of a carcinogenic food additive superfluous. For any agency 
that is permitted, or instructed, to weigh the costs and the health consequences 
of alternative means of controlling exposure, however, another solution is nec- 

The appropriate use of quantitative risk assessment has provoked fierce 
debate. Industrial interests have generally urged agencies to regard quantification 
as an essential step in evaluating measures for controlling carcinogens. -^^"^ On 
the other hand, consumer groups, labor unions, and some members of the public 
health community have cautioned against excessive reliance on techniques whose 
reliability remains uncertain. "^^"^ The regulatory agencies themselves for some 
time had difficulty reaching agreement on the proper role of quantitative risk 
assessment. -^^^ 

Opposition to quantitative risk assessment stems in part from doubts about 
its reliability. Critics of the technique also point to the wide range or risks at 

321. See Lave, supra note 286, at 35-36. 

322. Id. 

323. E.g., American Industrial Health Council, Proposal for a Science Panel (1980); GMA 
Petition, supra, at 95-130. 

324. See Letter from Arthur Upton to FDA Commissioner Donald Kennedy, October 1978. 

325. OSHA Carcinogen Policy at 5196-97. "In summary, EPA is the only one of the five 
agencies whose procedures were discussed in the record which makes quantitative estimates of risk 
on a regular basis in determining the degree of regulation." Compare IRLG Scientific Principles at 


low doses predicted by the different extrapolation models. ^^^ Furthermore, they 
recognize that quantitative assessment of risk facilitates assigning a dollar value 
to human life. Describing the health benefits of regulation in terms of the number 
of cancers that could otherwise occur permits calculation of a monetary value 
for each unit of risk. Regulatory agencies have studiously avoided this calcu- 
lation, because they fear that it will be misunderstood and question whether it 
would aid decisionmaking. -^^^ 

Despite these criticisms, quantitative risk assessment can help illuminate 
many of the choices that agencies confront in regulating carcinogens. For ex- 
ample, the technique facilitates comparison of the risks posed by different sub- 
stances, which aid in establishing priorities for regulation and permit selection 
of control levels based on their consistency with prior decisions. ^^^ Similarly, 
quantitative risk assessment can illuminate comparisons among regulatory op- 
tions. For example, to control exposure to a carcinogenic pesticide EPA has 
several options, including requiring that the pesticide be available only to profes- 
sional applicators, ordering protective safeguards for such individuals, restricting 
the crops to which it can be applied or areas in which it may be used, and 
specifying processing of treated foods. A common metric for measuring the 
health benefits of these various options can materially advance analysis of a 
multidimensional decision even if the measurements are readily verifiable and 
the benefits are not converted into dollars. Perhaps not surprisingly, EPA has 
been the chief proponent of quantitative risk assessment among the agencies. ^^^ 

For some time, FDA has also been an advocate of quantification. It was 
the first agency to propose basing permissible human exposure levels for car- 
cinogens primarily on quantitative extrapolation from high dose effects in lab- 
oratory animals. This is the thrust of FDA's proposed criteria for the evaluation 
of carcinogenic animal drugs. ''-^^ The agency has similarly used quantitative risk 
assessment to explain its decision to approve a carcinogenic hair dye ingredient, 
concluding that any risk of cancer associated with amounts of the dye that might 
be absorbed would be negligible. ^^' 

The CPSC's attitude to quantitative risk assessment has fluctuated. While 
the agency's staff performed quantitative assessments of the risk posed by several 
of the carcinogens it first regulated, the Commissioners conspicuously refrained 
from relying on these estimates to support their decision. In its 1978 cancer 
policy, subsequently withdrawn, the agency skirted the issue. ^^^ Later, in an 
advance notice of proposed rulemaking the Commission announced that it would 

326. See note 308 and Table III supra; OSHA Carcinogen Policy at 5200-OL 

327. But see 46 Fed. Reg. Ill 88, 1 1201-02 (1981), in which the CPSC depicts both dollar 
costs and estimated lives saved from ban on UF foam insulation. 

328. OSHA Carcinogen Policy at 5200-01 . 

329. See, text accompanying note 44 infra, Part II. But see NAS, Regulating Pesticides 
(1980), which criticizes EPA's reliance on quantitative extrapolation from animal studies to estimate 
human risk. 

330. See notes 98-100 supra and accompanying text. 

331. See note 97 supra and accompanying text. 

332. 43 Fed. Reg. 25658 (1978); see Merrill, Consumer Products at 1297-1300. 


seek to eliminate all "non-essential" uses of asbestos in consumer products, 
regardless of the associated level of risk.^^^ One year later, in proposing to ban 
urea formaldehyde (UF) foam insulation because of the cancer risk associated 
with formaldehyde fumes, the agency referred approvingly to a quantitative risk 
assessment performed by its staff. ^^"^ These contrasting approaches may simply 
reflect the Commission's anticipation of a potential problem: quantitative risk 
assessment requires good data about both carcinogenic potency and human ex- 
posure. The CPSC's asbestos notice contemplated a proceeding that could result 
in a generic ban on the use of asbestos in many consumer products. ^-^^ To make 
such an initiative dependent on determining the magnitude of the risk of each 
type of product could stall regulation indefinitely. 

For most of its existence OSHA was openly reluctant to rely on quantitative 
risk assessment in setting exposure limits for occupational carcinogens. ^^^ This 
reluctance ostensibly stemmed from the agency's interpretation of its statutory 
duty to eliminate, "to the extent feasible," exposure to substances that might 
harm even one worker. But there is another explanation for OSHA's position. 
Quantitative assessment can be used, implicitly or expressly, to assign a dollar 
value to human lives; OSHA has eschewed any technique that might appear to 
facilitate sacrificing members of its main political constituency, the labor unions. ^^^ 
In its 1980 carcinogen policy, OSHA declined to rely on quantitative risk as- 
sessment to set worker exposure levels but contemplated using it for two other 
purposes: ranking candidates for regulation, and illustrating the residual risk 
posed by exposure to a substance following implementation of mandated con- 
trols. ^^^ 

Even if the results of the 1980 election had not altered OSHA's direction, 
it is doubtful whether the agency could have continued to eschew quantification 
in setting health standards for carcinogens. The Supreme Court's decision in the 
benzene case^^^ would have forced some shift. Justice Stevens' opinion does 
not demand quantification where available data do not permit it; nonetheless, 
his insistence that OSHA must initially determine whether a carcinogen poses a 
"significant" risk requires that the agency find some way of describing the 
magnitude of the health consequences of prevailing worker exposures. The ex- 
trapolation procedures described above are the best developed techniques for 
doing this. 

How OSHA will use quantitative risk assessment is less certain. President 
Reagan's Executive Order directing executive branch agencies to weigh the costs 
and benefits of regulatory initiatives^'^^ would ostensibly necessitate some quan- 

333. 44 Fed. Reg. 60056 (1979); Merrill, Consumer Products at 1351-54. 

334. 46 Fed. Reg. 1 1 188 (1981); Merrill, Consumer Products at 1359. 

335. See id. at 1352-53. 

336. OSHA Draft at 95-98. 

337. Interview with Anson Keller, Esq., former Special Assistant to the Assistant Secretary 
of Labor, August 1980. 

338. OSHA Generic Carcinogen Policy at 5200-01 . 

339. Industrial Union Dep't, AFL-CIO v. American Petroleum Inst., 448 U.S. 607 (1980). 

340. Exec. Order No. 12291, 46 Fed. Reg. 13193 (1981). 


tification of health effects. But the Supreme Court's refusal to interpret the OSHA 
Act as requiring cost-benefit analysis^"^' dilutes this directive. Labor represen- 
tatives will argue that the Act's "feasibility" standard, as now construed, pre- 
cludes balancing, and that quantification is therefore unnecessary in setting exposure 
limits for carcinogens. On the other hand, the Administration's Task Force on 
Regulatory Relief may contend that the President can require his subordinates 
to quantify costs and benefits even if they are not permitted by statute to im- 
plement the decisions that such analysis supports. ^^^ Furthermore, OSHA's new 
Administrator has been quoted as advocating the use of cost-effectiveness anal- 
ysis to determine the measures to achieve compliance with desired exposure 
levels. This approach, if implemented, could also lead the agency to use quan- 
titative risk assessment in setting health standards. 

F. Regulatory Assessment of Costs of Controlling Exposure 

1 . Obligation to Consider Costs. 

As observed previously, some statutes do not require an agency to consider 
any economic consequences before mandating limits on human exposure to a 
chemical; the agency need consider only whether the substance presents a risk 
of cancer and whether humans are exposed to it.^"^-^ An agency using a negligible 
risk standard must go one step further and estimate the magnitude of the risk.^"^"* 
Most statutes, however, require that the agency investigate at least some of the 
consequences of curtailing exposure. The several statutes considered in this study 
require different effects to be considered and demand varying degrees of analytic 
rigor. None of them explicitly mandates formal cost-benefit analysis and, to a 
considerable extent, the scope of the inquiry actually undertaken is within the 
control of each agency. 

Agency statutes are not the only source of directives to consider the costs 
of regulation. Both the Ford and the Carter Administrations established programs 
that required executive branch agencies to identify and evaluate certain conse- 
quences of their actions. -^"^^ Both programs were concerned with the economic 
effects of regulatory actions, although the Ford program grew out of an earlier 
initiative that ostensibly reflected concern for the natural environment as well. 
President Reagan has issued his own instructions to agencies to consider the 
economic effects that may flow from their actions. -^"^^ This latest Executive Order 
differs from its predecessors in two important respects. First, it requires agencies 

341. American Textile Mfrs. Inst. v. Marshall, 452 U.S. 490 (1981). 

342. For a skeptical discussion of the authority of the President to require compliance with 
the requirements of the Reagan Executive Order, see M. Rosenberg, Beyond the Limits of Executive 
Power: Presidential Control of Agency Rulemaking Under Executive Order 12291, 80 Mich. L. 
Rev. 193 (1981). 

343. See text accompanying notes 80-81 supra. 

344. See text accompanying notes 94-108 supra. 

345. See generally Rosenberg, Beyond the Limits of Executive Power: Presidential Control of 
Agency Rulemaking Under Executive Order 12,291 , 80 Mich. L. Rev. 193 ( 198 1 ); Bruff, Presidential 
Power and Administrative Rulemaking, 88 Yale L.J. 451 (1979). 

346. Exec. Order No. 12291, 46 Fed. Reg. 13193 (1981). 


to submit proposed and final regulations, accompanied by their economic anal- 
yses, to the Office of Management and Budget prior to publication, thus affording 
the president's advisers an opportunity to influence their content. "^"^^ Second, it 
expressly provides that, to the extent compatible with other legal requirements, 
regulatory actions are not to be taken unless they meet a cost-benefit test.^"^^ 
This latter requirement may ultimately prove more influential than the typically 
more general language of Congress. 

2. Premises of Cost-Benefit Analysis. 

Two premises underlie the demand for systematic analysis of the effects of 
regulatory action as an integral part of decisionmaking. The first is the prosaic 
belief that rational choicemaking ought to be based on full awareness of the 
consequences. The second is the now vivid awareness, long ago illuminated by 
the discipline of economics, that society's resources are limited, and that the 
allocation of a portion of those resources to one objective necessarily displaces 
others. This perception is implicit in statutory directives to control risks that are 
"unreasonable" or to take into account specified collateral effects of regula- 

Economic techniques and decision analysis have been used to help make, 
or to evaluate, governmental decisions for several years. Methods of analysis 
that originated in the Defense Department during World War II have more 
recently been employed in such diverse areas as evaluation of water control and 
reclamation projects, regulation of foreign trade, and assessments of a broad 
range of social action programs. ^^^ Cost-benefit analysis is only one method for 
systematic analysis of alternative courses of action. In its most general form, 
cost-benefit analysis entails an accounting of all of the positive consequences 
that can be expected to result from a decision or program, an accounting of all 
the anticipated negative consequences, and a comparison of these consequences 
(i.e., costs and benefits) along chosen criteria of measurement. 

The theoretical rationale for using cost-benefit analysis in making regulatory 
decisions can be easily summarized briefly. Government regulation exists for 
the purpose of producing certain desired objectives or preventing certain evils. 
Many regulatory programs are directed toward stated or understood dominant 
goals, e.g., protecting consumers from cancer-causing agents in the food supply. 
If this were society's only goal, any action that reduced the risk of cancer from 
food would be desirable. But preventing cancer is only one of many societal 
objectives. Other goals include providing adequate sources of nutrition at a 
reasonable cost, preventing other food-borne diseases such as botulism, and 
fostering consumption of products that satisfy consumers. Actions designed to 

347. Id. 

348. Id. at 13193-94. 

349. See text accompanying notes 109-31 supra. 

350. See generally Tribe, Policy Science: Analysis or Ideology? in Philosophy and Public Affairs 
67 (October 1972); see also Report of the Social and Economic Committee of the Food Safety 
Council, Principles and Processes for Making Food Safety Decisions, Chapter 1 (Tent, draft, 
December 13, 1979). 


prevent cancer may often impede achievement of these other goals. An action, 
therefore, should not be evaluated solely in terms of the goal that it furthers, 
but also in terms of the goals that it sacrifices. By systematically analyzing the 
costs and benefits of a given course of action, a decisionmaker seeks to determine 
whether the net achievement of goals is positive. 

Cost-benefit analysis is distinguishable from cost-effectiveness analysis. 
Both share the assumption that a decisionmaker should take into account all of 
the effects of decisions, but cost-effectiveness analysis makes more limited claims. 
It does not purport to help the decisionmaker determine whether a goal is worth 
pursuing; rather, it assists him to identify the least costly way of achieving goals 
that have already been established. ^^^ For example, if it were accepted that 
worker exposure to benzene should be limited to 1 ppm, cost-effectiveness 
analysis could reveal what combination of methods, e.g., engineering controls, 
formulation changes, and worker protective equipment, will achieve this level 
most economically. ^^^ Cost-benefit analysis, by contrast, theoretically purports 
to identify which goals are, in economic terms, worth pursuing. 

The term "cost-benefit analysis" as used in this study embraces a decisional 
approach that is sometimes referred to as "risk-benefit analysis. "^^^ Cost-benefit 
analysis is potentially applicable to a much broader range of decisions than those 
in which the main variables are risk to human health on the one hand, and the 
costs controlling that risk on the other. Risk-benefit analysis can be viewed as 
a special variety of cost-benefit analysis, applicable, as the term applies, when 
one of the variables is the possiblity of some harmful event. ^^"^ To avoid possible 
confusion, only cost-benefit analysis is used here. Within this formulation "cost" 
will embrace all of the values resulting from the use of or exposure to a carcin- 
ogen, including satisfaction likely to be foregone and expenditures necessary to 
control exposure. "Benefit" will include the health benefits associated with 
reducing exposure, i.e., the health gain likely to be achieved by limiting or 
eliminating risk. 

3. Cost-Benefit Analysis in Health Regulation. 

a. Uses of cost-benefit analysis. Rigorous cost-benefit analysis requires 
the decisionmaker to identify all of the outcomes of a contemplated action, assess 
their likelihood and magnitude, and convert them to a common measure — 
typically dollars — to determine whether benifits outweigh the costs. ^^^ Obviously 
this theoretical ideal is routinely compromised, even in less complex contexts. 
The inability to anticipate consequences, the lack or cost of information about 
anticipated effects, and the difficulty of measuring some effects and converting 
them into a common metric routinely force decisionmakers to settle for much 

35 L See Lave, The Strategy of Social Regulation, 19-21 (1981). 

352. Compare id. at 90-91. 

353. This approach merges two types of analysis that others have seen as distinct. See id. at 
17-19; 23-25. 

354. See generally Crouch and Wilson, Risk/Benefit Analysis (1982); Lave, supra, at 23-24. 

355. See id. at 23-25. 


less. Other constraints inhibit government regulators from fully exploiting cost- 
benefit analysis. ^^^ Despite these limitations, however, the technique can be 

Cost-benefit analysis can help describe regulatory decisions. Any action by 
government trades off costs and benefits, i.e., any action has effects that are 
not desired even though they may not be large or of official concern. For example, 
a decision by FDA not to approve a new color additive has the effect, among 
others, of depriving consumers of another hue in their food — not to mention 
depriving the manufacturer of a potential market. Thus even a decision that 
purports to rest on the single criterion of risk in broader sense represents a trade- 
off between risk avoidance and other values. Anticipated effects are often ig- 
nored, either because they are expensive or even embarrassing to assess, or 
because they are beyond the jurisdiction of the decisionmaker to consider. ^^^ 
Some effects remain unexplored because they are not recognized or because no 
one has the incentive or the means to investigate them. 

Cost-benefit analysis can be useful in evaluating decisions made under less 
rigorous criteria. It provides one framework for critiquing government perfor- 
mance, as it has done for proponents of regulatory reform. ^^^ Cost-benefit anal- 
ysis can be used to assess not only specific actions, e.g., the decision to reduce 
benzene exposure from 10 to 1 ppm, but also the appropriateness of a series of 
decisions, e.g., all EPA attempts to control carcinogenic pesticides. ^^^ 

Finally, many statutes require agencies to make trade-offs each time they 
decide whether and how to regulate. For example, EPA must evaluate each 
pesticide under the "unreasonable adverse effects" standard, ^^ most of the 
economic costs and the health benefits of limiting exposure to a substance. It is 
for such individual decisions — to approve, ban, or restrict a substance — that 
this analytical technique is claimed to be particularly useful. 

b. Agency consideration of costs. None of the four agencies studied here 
has used classical cost-benefit analysis in making decisions. ^^^ EPA's approach 
to regulating carcinogenic pesticides most closely approaches the theoretical 
model, but still deviates in important specifics. The agency does not pretend to 

356. Shortage of time and resources to investigate possible effects that are anticipated explain 
the failure of many decisionmakers to explore fully all of the ramifications of a decision. For example, 
FDA has refused to consider the so-called "secondary economic effects" of alternative tolerances 
for food contaminants, such as the impact on sellers of sport fishing equipment of a reduction in 
the tolerance for PCBs in fresh-water fish. See Merrill and Schewel, Environmental Contaminants 
at 1415-23. Inadequacies of agency staff sometimes also limit inquiry. See id. at 1425. 

357. For example, in proposing to ban saccharin from food in 1977, FDA did not attempt to 
identify or measure the economic consequences of such action. See Saccharin and its Salts, 42 Fed. 
Reg. 19996(1977). 

358. See, e.g., Morrall, OSHA After Ten Years (paper prepared for AEI Conference on Health, 
Safety, and Environmental Regulation, November 1981). 

359. One of the classic studies of this type was Professor Peltzman's evaluation of the impact 
of FDA's efforts to implement the drug effectiveness requirements of the 1962 Drug Amendments. 
Peltzman, Regulation of Pharmaceutical Innovation (1974). 

360. See text accompanying notes 122-31, 166-84 supra. 

361. See text accompanying notes 83-87 supra. 


consider all possible consequences of its actions, and it has refrained from 
attempting to value health and environmental effects in dollar terms. Each of 
the other three considers a narrower range of costs. However, none of the agencies 
ignores the costs of restricting exposure to carcinogens. 

Costs are considered in many ways. Such consideration may be reflected 
in an agency's decision not to investigate a chemical, to postpone action and 
conduct another test, or to request further evaluation of health effects. Sometimes 
costs are discussed casually in an agency decision that is not subject to effect 
challenge. ^^^ Increasingly they are influential because participants in regulatory 
proceedings, often the agency itself as well as private parties, make substantial 
efforts to estimate their probability and magnitude. An agency's public statements 
are an unreliable guide to the weight actually given to costs. FDA's announcement 
in 1972 that it was again asking the National Academy of Sciences to assess 
saccharin's carcinogenicity did not mention the decisive consideration that sac- 
charin was the last non-nutritive sweetener. ^^^ Moreover, even when an agency 
purports to weigh a wide range of costs as well as possible health benefits, one 
cannot be sure whether the analysis dictated its choice or simply provided a 
rationale for an intuitive judgment. 

It should come as no surprise that agencies in fact consider consequences 
which they do not mention and sometimes mention consequences that they do 
not seriously consider. The health regulatory agencies operate at the intersection 
between the historical precepts of administrative law and the realities of con- 
temporary government. Much legal scholarship and many judicial decisions 
support the assumption that regulation is the exercise of expertise applied to facts 
whose relevance is legislatively determined. The realities of course, are that 
regulators are selected as much for their predilections as their expertise, that 
they must make decisions when many facts are unknown and — in the field of 
immediate interest — unknowable, and that they make discretionary choices be- 
tween competing values. ^^"^ There thus is a basic tension between the dictates 
of cost-benefit analysis and the institutional pressures of governing, and agencies 
vary both in their fidelity to and in their facility with systematic tools for de- 

4. Methodology of Cost-Benefit Analysis. 

a. Determining what costs are relevant. A threshold issue for any agency 
responsible for deciding whether to limit chemical exposures is what effects to 
consider. A major action, such as OSHA's coke oven emissions standard, ^^^ can 
have a broad range of short-term and long-term effects throughout the economy. 

362. For example, the CPSC's negotiated recall of hairdryers containing asbestos linings was 
accompanied by very little discussion of the economic effects of its action. See Merrill, Consumer 
Products at 1346-51. 

363. See Merrill, Saccharin: A Regulator's View, in Crandall and Lave, eds.. The Scientific 
Basis of Health and Safety Regulation 156-60 (1981). 

364. For provocative analyses of the purposes and performance of regulation, see Breyer, 
Reforming Regulation (1982); Noll, Reforming Regulation: An Evaluation of the Ash Council 
Proposals (1971). 

365. See Lave, The Strategy of Social Regulation, 93-98 (1981). 


An effort to identify all of the effects and estimate their magnitude would be an 
endless exercise some selectivity is accordingly essential if decisions are to be 
reached. Immediate guidance can often be found in an agency's statute. Congress 
may have specified the kinds of effects that it wanted the agency to consider, 
as it did in the Consumer Product Safety Act.^^^ Or, as in FIFRA, it may have 
provided a general standard for decision, together with obviously illustrative 
factors believed pertinent to the general inquiry. ^^^ Congress' failure to mention 
criteria that logically seem relevant may betray a decision that they should not 
be considered. ^^^ Presidential directives may not only mandate general consid- 
eration of regulatory costs but specify which costs are of particular interest. 
Executive Order No. 12291 enumerates several possible effects of regulatory 
actions, but they cover so wide a range that it is difficult to identify a category 
that an agency could confidently exclude. 

The Reagan order raises an unresolved issue about the seriousness with 
which agencies will investigate costs and benefits. It recognizes that some agen- 
cies may be constrained by law from according decisional significance to effects 
that such analysis might re veal. ^^^ The question is whether an agency that cannot 
lawfully take into account some, or any, costs of regulation must nonetheless 
estimate them. Would FDA now be obligated to attempt to measure the effects 
of banning saccharin on dieters and diabetics even though it may not lawfully 
consider them? The answer may simply be a matter of intent; arguably the 
President can require an official in the executive branch to perform any kind of 
study he considers useful. The question is not merely academic; investigating 
regulatory costs can itself be expensive of time and resources, and the obligation 
to perform such agency investigation can therefore inhibit actions as to which 
costs are not technically relevant. ^^^ 

While this discussion has focused on the negative consequences of restricting 
human exposure to chemicals, it is appropriate to point out that the benefit side 
of the equation also embraces a wide range of effects. This study concerns 
decisionmaking about chemicals whose primary adverse effect is cancer. The 
specific cases examined involve attempts to estimate the likelihood that exposure 
to a chemical will cause cancer in humans. But many agents capable of inducing 
chronic disease can also cause acute toxic effects. In the typical case the chemical 
is used or occurs at levels well below toxic doses, but this is not invariably true. 
Pesticides, for example, often must be used at levels that may pose acute risks 

366. See Merrill, Consumer Products at 1279-80. 

367. See text accompanying notes 166-79 supra. 

368. It seems highly probable that Congress's failure to instruct FDA to evaluate the utility or 
benefits of food additives reflected a conscious choice to exclude that criterion from the agency's 
purview. See Merrill, Regulating Carcinogens in Food: A Legislator' s Guide to the Food Safety 
Provisions of the Federal Food, Drug, and Cosmetic Act, 11 Mich. L. Rev. 171, 204-09 (1978). 

369. Exec. Order No. 12291 provides that "to the extent permitted by law," agencies shall 
adopt or repeal rules to maximize net benefits to society and that "regulatory action shall not be 
undertaken unless the potential benefits to society . . . outweigh the potential costs to soci- 
ety .. . Id., par. 2(b)-(c). 

370. It is possible, of course, that this is one of the objectives of the Reagan executive order. 


for applicators. Furthermore, some chemicals are capable of causing other chronic 
effects, e.g., genetic damage, or congenital deformity. Methods for identifying 
agents capable of causing such effects or estimating their frequency are far more 
primitive than the tools for evaluating carcinogens. ^^^ 

It should also be noted that some chemicals that pose a health risk also 
provide health benefits. A much-disputed illustration is sodium nitrite, a sub- 
stance used to cure meat in order, it is asserted, to prevent the formation of 
botulinal toxin but one that also may cause tumors in experimental animals. ^^^ 
There surely are health benefits directly associated with many rodenticides, not 
to mention the indirect benefits of more abundant, less costly food resulting from 
pesticide use. Such health benefits are to be distinguished from the reduction in 
cancer associated with restricting human exposure to chemicals; they belong on 
the same side of the equation as the costs of regulation. ^^^ 

b. Estimating costs. Under most statutes a central concern is the direct 
costs of restricting exposure to a carcinogen. Because agency programs provide 
different kinds of opportunities to curtail exposure, the character of these costs 
cannot be easily categorized. For example, given the practical impossibility of 
removing PCBs from fish, the immediate costs of FDA efforts to curtail expo- 
sure — by whatever means — will be measured by diminished sales of fish.^^"^ 
For OSHA, by contrast, the primary direct costs of control include expenditures 
for new equipment, process changes, and protective gear for workers. 

Before an agency can estimate such costs, it must consider what methods 
of controlling exposure are realistic options. In theory, many different methods 
could be used to reduce human exposure to a toxic chemical. These range from 
mandated information for consumers to process changes, engineering controls, 
personal protective devices, restrictions on the manner and frequency of use, 
and prohibition of the use or sale of the substance. Failure to investigate alter- 
native approaches could distort an agency's cost estimates, but some apparently 
sensible control methods may be beyond an agency's authority to implement. 
FDA, for example, has no option but to ban carcinogenic food additives. ^^^ 
Ostensibly its only means of controlling food contaminants is to condemn over- 
tolerance shipments of food, but presumably it could combine this approach with 
required label warnings against excessive consumption. ^^^ Rationing the food, 
however, would be beyond its authority. On the other hand, EPA may draw on 

371. See generally, NAS, Identifying and Estimating the Genetic Impact of Chemical Envi- 
ronmental Mutagens (1982). 

372. See NAS, The Health Effects of Nitrate, Nitrite, and N-Nitroso Compounds (1981). For 
such agents, a decisionmaking formula that allowed regulators to balance risks might be appropriate. 
See Lave, The Strategy of Social Regulation 15-16, 48-57 (1981). 

373. See id. 

374. See Merrill and Schewel, Environmental Contaminants at 1415-23. 

375. See text accompanying notes 88-95 supra. 

376. FDA has rarely used such an approach in regulating harmful agents in food, though the 
legislation that suspended FDA's authority to ban saccharin mandated a label warning on all saccharin- 
sweetened foods. See Merrill, Saccharin: A Regulator's View, supra, at 160. And public health 
authorities in other countries rely heavily on public information techniques in attempting to control 
the adverse health effects of food contaminants. 


several control options to reduce exposure to a carcinogenic pesticide, including 
use information, regional and crop limitations, use restrictions (e.g., by profes- 
sional applicators and cancellation i.e., banning). ^^^ A comparable range of 
approaches would be available to OSHA, but that agency historically has pre- 
ferred engineering controls over all other techniques for limiting worker exposure 
to toxic agents. ^^^ Other remedies sometimes suggested, such as taxes or other 
pecuniary incentives, are not currently available to any of the four agencies 
studied. The significance of this and similar limitations is that an agency will 
not devote much attention to control options that it cannot implement. 

Once an agency has identified the alternative control measures it will con- 
sider, it attempts to determine what the costs of each are likely to be. A full- 
fledged engineering and economic analysis of any option is time-consuming and 
expensive. Agencies are invariably short of funds to pay for studies and often 
lack personnel who understand the technology involved. These circumstances 
provide an incentive for agencies to narrow the range of options considered. ^^^ 
None of the agencies studied has authority to mandate the necessary studies by 
firms responsible for a chemical, but the prospect that an agency can be persuaded 
to choose, or must choose, ^^^ a less costly option sometimes elicits such studies. 
Where compliance is technologically complicated, private businesses and trade 
groups are the primary sources of data about control costs. Agencies routinely 
permit the submission of such data, and then have the difficult task of evaluating 
their reliability.^^' 

Process changes, engineering controls, warning labels, and personal pro- 
tective equipment all involve initial investments, most entail recurrent expense, 
and some may impose inconvenience, e.g., protective devices for workers. Many 
investments, however, serve more than one purpose, which should lead analysts 
to allocate estimated costs between compliance with regulations and the other 
purposes they serve. "^^^ Another difficulty in estimating costs stems from the 
inability to measure technological improvements that are likely to follow im- 
position of more stringent standards. Estimates based on present information and 
current technologies will be only rough approximations of what can be expected 
if new controls are imposed. ^^^ Thus, not only are cost estimates uncertain, but 
they are likely to be overestimates for they typically assume that innovation will 
not yield economies. ^^"^ Furthermore, many exposure controls have resulted in 

377. See NAS Regulating Pesticides 131-53 (1980). 

378. See OSHA Carcinogen Policy at 5222-27. 

379. See, e.g.. Lave, The Strategy of Social Regulation 129-35 (I98I). 

380. President Reagen's executive order would require agencies to the extent permitted by law, 
to choose the regulatory option that imposed the lowest costs on society. See Exec. Order No. 
12291, par. 2, 46 Fed. Reg. 13193 (1981). 

381. For an illustration of the difficulty of analyzing information about economic effects of 
regulation, see the discussion of OSHA's assessment of the feasibility of its cotton dust standard in 
American Textile Mfgrs. Ass'n v. Donovan, 452 U.S. 490 (1981). 

382. See Lave, The Strategy of Social Regulation, 36-38 (1981). 
383. Id. 

384. Id. 


savings of raw materials or finished product which would otherwise have been 
vented. ^^^ 

In addition to evaluating the direct costs of control options, an agency in 
theory ought to consider the cost of doing without the final product or paying 
more for it and the cost of substituting another substance. The first is particularly 
difficult to estimate, but banning may be the preferred or only lawful means for 
controlling a substance. One can approach this problem either by estimating the 
cost of a substitute product which consumers deem equivalent or by estimating 
how much consumers would be willing to pay to have the product rather than 
have it banned. ^^^ 

Other indirect costs of regulation may also merit consideration. Substances 
likely to be substituted for the restricted chemical may themselves entail health 
risks. For example, several years ago, under pressure from public health experts 
and regulators, dry cleaning establishments switched to perchloroethylene as 
their primary cleaning agent, only to learn in 1977 that the CPSC was evaluating 
this material as a potential carcinogen. ^^"^ With disconcerting frequency, agencies 
discover that less is known about the toxicity of possible substitutes than about 
the substance scheduled for regulation. ^^^ Judgments about the risk of such 
materials frequently reflect wishful thinking more than careful evaluation of 
experimental data.-^^^ 

The range of potentially costly secondary effects of regulation is extremely 
broad. They may include effects on industrial concentration, competition, un- 
employment compensation expenditures, and local tax revenues. One effect 
predicted to flow from FDA's adoption of a more stringent tolerance for PCB's 
in commercial fish is a decline in recreational fishing, with attendant losses for 
bait sellers, motel operators, and others who earn money from supplying va- 
cationers. ^^° Even though FDA's tolerance would not technically apply to fish 
taken by sport fishermen for their own use, it is difficult to deny — and even 
more difficult to measure — such secondary effects. ^^' 

c. Problems with cost-benefit analysis. A striking characteristic of deci- 
sions by the four agencies is the extent to which their analyses of the implications 
of specific actions depart from the economist's model. These departures are 
partly a function of statutory directives that appear, or have been construed, to 
limit the criteria regulators must consider. ^^^ But practical as well as theoretical 
problems with cost-benefit analysis have also discouraged its rigorous use. 

385. Id. 

386. Id. at 39. 

387. See Merrill, Consumer Products at 1297-1304. 

388. The fabric treatment TRIS came into wide use after the government had directed fabric 
makers to assure their products were flame retardent. See Merrill, Consumer Products at 1323. The 
CPSC also considered the potential toxicity of alternatives to benzene in such consumer products 
as rubber cement. See id. at 1343-44. 

389. Id. 

390. See Merrill and Schewel, Environmental Contaminants at 1415-23. 

391. FDA's position, however, is that it will not attempt to evaluate such effects in setting 
tolerances or action levels. See text accompanying notes 120-121 supra. 

392. See text accompanying notes 85-86 supra; OTA Report at 176-83. 


It is acknowledged that in the practical world many decisions do not require, 
or permit, the kind of careful investigation and quantification of effects that the 
cost-benefit model dictates. When an agency must act quickly or forfeit any 
opportunity to intervene, elaborate inquiry is not feasible. When the danger 
apprehended from a substance is very large, a decision is likely to rest on intuitive 
judgment. Furthermore, when the persons at risk are identifiable individuals, 
government rarely pauses to ask whether resources should be spent to protect 
them.^^^ But in regulatory contexts the individuals at risk are rarely readily 
identifiable i^^"^ the trade-off that agencies typically confront is statistical lives 
for dollars that can be used for other societal purposes. Even in such cases, 
however, regulators have often eschewed systematic analysis of costs and ben- 

One obvious explanation for this anomalous behavior is the expense of 
formal analysis. Information about costs and benefits is invariably incomplete. 
Obtaining more inevitably requires money and, as important, time. These are 
major incentives for both agencies and those affected by their decisions to accept 
short-cuts and approximations in place of accurate answers. The delay produced 
by requirements for rigorous analysis — to support more efficient regulation — is 
dual-edged. When an agency is confronted with a substance to which humans 
are already exposed, such a requirement impedes the adoption of new controls; 
when approval for a new product is being sought, however, insistence that the 
agency be satisfied that the benefits outweigh the costs slows the introduction 
of new technologies. ^^^ 

The administrative process further amplifies these costs of cost-benefit anal- 
ysis in a way many economists ignore. Regulation is not only a search for right 
answers; it is also an exercise in participatory government. The notion that 
affected individuals and organizations are entitled to "be heard" on proposals 
that may adversely affect their interests is deeply embedded in American law. 
Some judicial rulings have relaxed this principle: according to the Supreme Court, 
the requirement for a "hearing" can be satisfied by affording an opportunity 
for the submission of written data and arguments. ^^^ But other cases have steadily 
enlarged opportunities for interest group participation in agency decisionmaking. 
More dramatic has been the growth in complexity of the issues given to regulatory 
agencies to decide. The result is that any major controversy — and most decisions 
involving carcinogens fall in this category — can require protracted, expensive 

393. It is improbable that anyone performed a cost-benefit analysis of the Carter Administration's 
efforts to rescue the hostages from Iran before it was undertaken. And similar rescue efforts directed 
at saving individuals trapped in the mine explosions or imperiled by natural disasters would probably 
fail a cost-benefit test. 

394. Not only are the numbers of persons at risk often very large, even when the population 
at risk can be identified— as in occupational settings — those who will be "saved" by investments 
in more stringent safeguards are never known because individual susceptibility usually cannot be 
determined in advance. Furthermore, in many instances regulators cannot say with certainty that 
any members of an exposed group will get cancer; they can predict only that some percentage might. 

395. See, e.g., Peltzman, Regulation of Pharmaceutical Innovation (1974). 

396. See United States v. Florida East Coast Ry., 410 U.S. 224 (1973). 


proceedings. -^^^ There will be disputes over what costs are relevant, whether 
efforts to measure them are competent and honest, and how large they are. The 
same is true, of course, for the health components of the equation. Each element 
added to the formula for decision is potentially a separate battleground. 

On a more prosaic level, the reliability of data and the soundness of estimates 
or predictions are persistent problems. It is convincingly argued that industry 
estimates of compliance costs are often exaggerated. ^^^ The temptation to over- 
statement is obvious. Moreover, as Lave points out, even the honest firm is 
confined to existing information and current technology. ^^^ According to some 
critics of cost-benefit analysis, decisionmakers tend to accord greater weight to 
effects that are readily quantified, such as the cost of engineering controls, and 
less to health effects, which are more elusive. "^^^ Uncertainty pervades both 
estimates of the compliance costs and assessments of risk. Where achievement 
of mandated limits is difficult to monitor or requires the coordinated advertent 
behavior of many individuals, regulatory determinations of appropriate exposure 
levels must be recognized as very fragile. 

Another recurrent problem is the valuation of qualities or effects that are 
not customarily measured in dollars. Prosaic examples include the appearance 
of food, the appeal of special diets, and the satisfaction of having "burning" 
embers in one's living room fireplace. "^^^ But by far the most controversial such 
issue is the propriety of assigning dollar values to human life."^^^ Rigorous cost- 
benefit analysis requires the decisionmaker to quantify all significant effects, 
convert them to a common measure, and then determine whether benefits exceed 
costs. "^^^ Economists do not shrink from assigning an explicit value to human 
lives, e.g., the lives likely to be saved by reducing exposure to a carcinogen. 
They note that any careful observer of regulatory actions can retrospectively 

397. For an analysis of duration and complexity of OSHA proceedings to set standards for 
occupational carcinogens, see McGarity, The Occupational Safety and Health Administration's 
Generic Carcinogen Policy: Rulemaking Under Scientific and Legal Uncertainty (unpublished manu- 
script prepared for the National Center for Administrative Justice, 1980). See also EPA, Pesticide 
Programs; Rules Governing Rebuttable Presumption Against Registration (RPAR) Proceedings; Rules 
of Practice Governing Hearings Under Section 6 of the Federal Insecticide, Fungicide, and Roden- 
ticide Act (FIFRA), 45 Fed. Reg. 52628 (1980) (proposing revisions in increasingly lengthy and 
costly procedures for pesticide cancellation). 

398. Perhaps the most frequently cited case involved OSHA's proceeding to reduce the exposure 
limit for vinyl chloride, in which it mandated steps that the industry contended would cost in excess 
of $2 billion and threaten bankruptcy. Once the standard became final and compliance was attempted, 
the actual expenditures has shrunk by more than half. See Doniger, Federal Regulation of Vinyl 
Chloride: A Short Course in the Law and Policy of Toxic Substances Control, 7 Ecology L.Q. 497, 
561 (1978). 

399. Lave, The Strategy of Social Regulation 36-38 (1981). 

400. See generally, Swartzman, Liroff, and Croke, eds., Cost-Benefit Analysis and Environ- 
mental Regulations, 53-85 (1982). 

401. See Lave, The Strategy of Social Regulation 39-41 (1981). See generally, Swartzman et 
al, supra, note 400. 

402. See, e.g., Kelman, Cost-Benefit Analysis— An Ethical Critique, Regulation, Jan./Feb. 
1981; Lave, supra, note 401 at 40. 

403. Id. at 23-25. 


compute the value agencies implicitly placed on lives. Several formulae have 
been proposed for formally determining what a life is worth /^"^ 

A few government agencies have explicitly assigned lives dollar values in 
allocating resources to health protection or accident prevention, "^^^ but most 
regulators have spumed the very notion. Even EPA, whose analyses of regulatory 
options have typically been the most comprehensive, has refrained from valuing 
lives at risk to compare with the costs of control. This reluctance has ethical as 
well as practical roots. Furthermore, the suggested criteria for computing the 
value of life, notably discounted future earnings and willingness to pay,"^^^ pro- 
duce widely divergent values, and no consensus has emerged in favor of one 
approach. There is also debate over whether future health benefits should be 
discounted, in the same fashion as future control expenditures. "^^^ 

There is a more fundamental problem with using formal cost-benefit analysis 
to set exposure limits for carcinogens. Even if an agency could determine that, 
in aggregate, the costs of an action outweighed the health benefits, this result 
would not necessarily dictate that the action should not be taken. None of the 
statutes examined here expresses a preference for economic efficiency as the 
prime determinant of regulatory policy. A carcinogen's risks are rarely borne 
only by those who derive the benefits, and who thus would bear the costs of its 
control. "^^^ Exposed workers, to be sure, benefit from jobs that involve the use 
or production of benzene, but substantial benefits are also enjoyed by other 
groups who bear none of the risks. There is similar lack of congruence between 
risk bearers and beneficiaries of consumer products, contaminated food, and 

Congress might legitimately wish specific agencies to give preferential weight 
to the risks borne by members of particular groups even if more than offset by 
the benefits to others. Such a preference, if apparent, could reflect a judgment 
that maldistributioin of wealth diminished their opportunity to bargain for pro- 
tection, or that the delayed effect of carcinogens makes it unrealistic to assume 
that the market will efficiently allocate the costs of preventing the disease. It is 
often difficult, of course, to discern what theory underlies regulatory legislation, 
but there is little basis for supposing that Congress wished the agencies invariably 
to strive for maximum efficiency. Indeed, there is good reason to believe that 
at least two of the statutes studied — the Occupational Safety and Health Act"^^^ 

404. See, e.g., Bailey, Reducing Risks to Life: Measurement of the Benefits, (AEI 1980). 

405. Id. at 26. 

406. See, e.g., Kelman, Cost-Benefit Analysis and Environmental, Safety, and Health Regu- 
lation: Ethical and Philosophical Considerations, in Swartzman et al, Cost-Benefit Analysis and 
Environmental Regulations 137-51 (1982). 

407. See Lave, The Strategy of Social Regulation 41-45 (1981). 

408. The congruence between beneficiaries and risk-bearers may be closer for toxic substances 
encountered in the food supply, such as saccharin; they are probably most remote for agents that 
disperse and persist in the environment. 

409. See, American Textile Mfgrs. Ass'n v. Donovan, 452 U.S. 490 (1981). 


and the Food, Drug, and Cosmetic Act'^'^ — embody an emphatic, if imprecise, 
Congressional preference for preventing harm even at substantial cost. 

This perception is surely shared in varying degrees by the four agencies 
studied. Agency employees have a highly-developed sense that their mission is 
health protection."^'' Until recently, this sense mirrored the risk-averse posture 
of a majority of Congress, at least as measured by support for new legisaltion 
and by oversight hearings that dramatized bureaucratic failures to restrict ex- 
posure to risky substances'^'-^ This atmosphere would have discouraged any agency 
from following a rigorous cost-benefit approach. Recent shifts in public percep- 
tion of the costs of regulation make such analysis more congenial, but it may 
also impede adherence to the economists' models. The popularity of "regulatory 
reform" stems only partly from systematic analysis of agency decisions. It 
reflects as well an untutored conviction that American industry has been exces- 
sively burdened by regulation, a conviction that will continue to exert pressure 
to hold down costs regardless of health benefits. 

410. See, text accompanying notes 88-94, supra; Merrill, Regulating Carcinogens in Food, 
supra, note 368, at 183. 

411. The best evidence of this commitment — or perhaps bias — is the outpouring of criticism 
of EPA actions and decisions from its own employees during the last 18 months. FDA Commissioner 
Alexander M. Schmidt encountered similar criticism from within the agency, which was dramatized 
during hearings before Senator Edward Kennedy's Administrative Practices and Procedures Sub- 
committee of the Senate Judiciary Committee in 1974 and 1975. See, Final Report of the Review 
Panel on New Drug Regulation (DHEW 1977). 

412. Here again the examples are numerous. One of the best illustrations of the emphatically 
risk-averse posture espoused by many members of Congress during the 1970's is the final report of 
the Moss Subcommittee investigation of the performance of nine regulatory agencies, including 
EPA, FDA, and the CPSC. Federal Regulation and Regulatory Reform, Report by the Subcommittee 
on Oversight and Investigations of the House Committee on Interstate and Foreign Commerce, 94th 
Cong., 2d Sess. (1976). The climate in Congress began to change following the 1976 election and 
perhaps was influenced by public reaction to FDA's proposed ban of saccharin. 


Part II: Procedures for Agency Decisionmaking 
A . Introduction 

1 . Scope and Organization of Part II. 

This part evaluates the existing legal and institutional mechanisms for reg- 
ulating chemical carcinogens and offers recommendations for their improvement. 
Some recommendations are directed at individual agencies, others speak to the 
four agencies collectively, and still others address decisionmaking at the inter- 
agency level. The part also discusses possible new institutions, such as the 
proposed National Science Council which would determine the risk of compounds 
considered for regulation by any agency.' 

The following discussion assumes some familiarity with the four agency 
programs studied. Part I of this report, 'The System for Regulating Chemical 
Carcinogens," describes the scientific and statutory context of toxic chemical 
regulation in the United States, but does not comprehensively treat each agency 
program, each of which is the subject of a separate part. Two of these have been 
published separately: Merrill and Schewel, FDA Regulation of Environmental 
Contaminants of Food, 66 Va. L. Rev. 1357 (1980), and Merrill, CPSC Reg- 
ulation of Cancer Risks in Consumer Products: 1972-1981 , 67 Va. L. Rev. 
1261 (1981). Drafts of the remaining two parts are available from the Admin- 
istrative Conference. It should be noted that several recommendations speak to 
procedures or institutions that are not discussed in earlier parts. Where recom- 
mendations go beyond the context previously established, this part provides 
sufficient background to enable the reader to evaluate them. 

Another introductory caveat is in order. To characterize the government's 
efforts to prevent human exposure to chemical carcinogens as controversial would 
be extreme understatement. Perhaps no facet of federal regulation occasioned 
as intense debate during the 1970's. Agency programs have been singled out as 
imposing excessive costs on the industrial sector,- and individual decisions have 
been challenged as indifferent to control costs. "* In other quarters, all four agencies 
have repeatedly been criticized for failure to prevent avoidable disease. Beyond 
polemics, debate has focused on government's proper role in health protection.'* 

While this report thus deals with agencies whose actions have been the 
targets of prolonged dispute, it does not address many of the underlying issues. 
To be sure, in discussing both specific decisions and general programs, the report 
inevitably suggests views about their appropriateness; its main concern, however, 
is with the procedures the agencies follow and the competence they display. The 

1. See National Science Council Act of 1980, H.R. 638, 97th Cong., 1st Sess, 127 Cong. 
Rec. 267 (1981). 

2. J. Morrall, OSHA After Ten Years, Preliminary Draft, Ch. 5, pp. 5-7; "Regulation Cost 
2.6 Billion for 48 Companies," Washington Post, March 15, 1979. 

3. Id. 

4. Nichols and Zeckhauser, Government Comes to the Workplace: An Assessment of OSHA, 
54 Pub. Int. 39, 55-57 (1979). 


report thus does not discuss the extent to which human exposure to carcinogens 
should be determined by command-and-control regulation rather than market 
forces. Nor does it endorse cost-benefit analysis as the criterion for regulating 
carcinogens. These obviously important issues are beyond its scope. 

At the outset of a necessarily lengthy discussion it is useful to outline the 
sequence in which topics are treated. Section B describes the common elements 
of government decisions to regulate a chemical. This description provides a 
framework for discussing recommended improvements in the regulatory process. 
The latter portion of Section B attempts to characterize these elements and 
suggests criteria for evaluating current agency performance and formulating 
proposals for improvement. Section C considers the means by which agencies 
identify candidates for regulation. Section D explores the methods they use in 
evaluating the health risks of toxic chemicals, and specifically considers the 
values and limits of quantitative risk assessment. Section E examines methods 
for generating information about, and evaluating, the costs of controlling ex- 
posure to carcinogens. It also explores the procedural and administrative impli- 
cations of cost-benefit decisional criteria. Section F treats issues arising during 
the formal administrative process, i.e., the stages of agency decisionmaking that 
are governed by statutorily described procedures, such as those required by the 
Administrative Procedure Act (APA).^ Finally, Section G discusses opportunities 
for, and obstacles to, interagency collaboration in regulating carcinogens. Rec- 
ommendations for improving agency performance appear throughout. 

B. Elements of Agency Decisionmaking 

This section establishes a framework for evaluating agency procedures for 
regulating carcinogens. The framework is a simplified decision sequence, which 
is based upon the experience of the four agencies studied. This prototype is not 
a description of administrative procedure, i.e., its elements do not correspond 
to stages of recognizable legal processes. Rather, it represents a series of findings 
that are implicit, and often articulate, in any government decision to regulate a 
chemical. In short, the prototype describes the thought process of government 
regulators. The formal procedures — rulemaking, adjudication, or licensing — 
required to implement an agency's analysis sometimes occur contemporaneously, 
but often they are not commenced until agency decisionmakers have tentatively 
resolved the issues described below. 

1 . Discovering Targets for Regulation . 

The health hazards that FDA, CPSC, OSHA, and EPA have considered 
for regulation have come to their attention by different routes. With some ov- 
ersimplification, these routes can be categorized under four headings: 

a. Report of laboratory test results. Most candidates for regulation have 
been identified through experimental findings of carcinogenicity. The CPSC's 
1980 proposed ban of UF foam insulation was predicated principally on the 

5. 5 U.S.C. §§551 etseq. (1976). 


results of one experiment, confirmed by a panel of federal scientists, indicating 
that formaldehyde is carcinogenic for rats and mice exposed by inhalation.^ 
Similarly, FDA's ban of chloroform as an ingredient in cosmetics and most drugs 
was precipitated by a positive bioassay report from the National Cancer Institute.^ 
The agency had neglected the possible health hazards of chloroform before the 
NCI study was reported; its decision to regulate the chemical was a direct response 
to the study, not the completion of an inquiry begun previously. 

This pattern of hurried response to unexpected experimental findings is no 
longer as common. Regulatory agencies are now generally aware of what chem- 
icals are under test by federal laboratories, and have increasing knowledge of 
testing being conducted within the industrial community as well.^ Many of the 
chemicals being tested under government auspices were nominated by one or 
more of the agencies.^ Experimental scientists will continue to produce surprises, 
however, and regulators cannot escape the responsibility — imposed by political 
pressure as well as by concern for public health — to respond to unanticipated 
disclosures of apparent health hazards.'^' The global market for chemicals and 
the intemationality of toxicological research make it impossible for agencies to 
monitor more than a small percentage of ongoing studies. 

b. Discovery of human exposures to, or illnesses caused by, a substance 
believed to be toxic. An agency may first become concerned about a toxic 
substance after some event reveals that humans are exposed to it. For example, 
FDA first discovered the hazard of polybrominated biphenyls (PBBs) when dairy 
farmers in Michigan began reporting unexplained illness among their cattle and 
later investigation identified PBBs as the responsible agent." FDA previously 
had no reason to think that the chemical might be a contaminant of food and 
thus had neither looked for it or investigated its toxicity. Reports of unusual 
human illness associated with exposure to a chemical may also precipitate reg- 
ulatory attention. OSHA's standard for vinyl chloride was largely a response to 
the disturbing discovery of an unmistakable association between workplace ex- 
posure and angiosarcoma, a cancer rarely found in the liver. '^ 

c. Applications for marketing approval. An agency's attention may be 
called to a chemical by an application for marketing approval or a notice of 
intention to distribute. Under regulatory programs that require proof of safety 
before a chemical can be marketed, an agency can demand considerable infor- 
mation concerning its potential health effects. The authority to require testing, 

6. 46 Fed. Reg. 11188 (1981). 

7. 41 Fed. Reg. 26845 (1976). 

8. The CPSC was advised about the results of the civil study of formaldehyde long before it 
was completed. 

9. National Toxicology Program, NTP Technical Bulletin 4: 3-6 (1981). 

10. An example is the discovery, by scientists at the Center for Disease Control, of a probable 
association between use of super absorbent tampons and toxic shock syndrome. 

11. Merrill and Schewel, Environmental Contaminants at 1410-1415. 

12. R. Merrill, Draft of Chapter 4: OSHA Regulation of Toxic Chemicals in the Workplace 
[hereafter OSHA Draft] at 145. 


coupled with the obstacles to approval of any chemical shown to cause cancer 
in animals, '"* has meant that few new carcinogens have come to agency attention 
through this route. Only substantial commercial promise will induce a firm to 
attempt to persuade FDA or EPA than an animal carcinogen should be approved 
for distribution even if the law theoretically allows it. 

These incentives have not meant, however, that chemicals requiring pre- 
market approval have not been regulated as carcinogens. The most controversial 
candidates for regulation have been approved compounds whose carcinogenicity 
was suggested long after commercial introduction. Saccharin, nitrite, cyclamate, 
DDT, heptachlor and chlordane, and 2,4, 5-T all illustrate this phenomenon. In 
most of these cases, administrative attention was attracted by the discovery of 
unexpected human exposures or the report of new laboratory findings.'"^ While 
at first glance perplexing, we should not be surprised when a substance previously 
approved as safe is discovered to be carcinogenic. Standard protocols for chronic 
testing are now much more thorough, and continued experimentation may expose 
species or genetic strains of animals that are more sensitive to a chemical than 
those originally tested.'^ 

d. Systematic review. The four agencies have often been criticized for 
their reactive posture toward chemical carcinogens. '^ Concern for public health 
and a desire to control their own agendas have inspired the agencies to devise 
systematic ways of reviewing chemicals within their jurisdiction in order to 
identify those that may warrant regulation. For at least a decade FDA has con- 
ducted systematic reviews of the safety, and in some cases the effectiveness, of 
substances used as ingredients in foods, OTC drugs, and biological products.'^ 
EPA has largely pursued its own agenda under the RPAR program for registered 
pesticides. ^^ During the 1970's the CPSC responded primarily to targets identified 
by outside bodies, but in 1979, it established an internal review system, focusing 
on exposures that might be hazardous as well as chemicals to which consumers 
might be exposed.'^ OSHA's generic carcinogen policy outlines a system for 

13. Very few animal carcinogens have been approved by government agencies when this 
information was available to them in advance. Even though the Delaney Clause has very narrow 
application, few manufacturers would persist in seeking marketing approval for a new substance 
shown to cause cancer in laboratory animals. 

14. Nitrite, saccharin, and cyclamates to mention only three, all became of regulatory interest 
as the result of reports of studies in laboratory animals. See R. Merrill, Regulating Carcinogens in 
Food: A Legislator' s Guide to the Food Safety Provisions of the Federal Food, Drug, and Cosmetic 
Act, 11 Mich. L. Rev. 171, 181 (1978) [hereafter Carcinogens in Food]. 

15. Correspondence to author from Richard Bates, M.D., Clement Associates, December 31, 
1981 [hereafter Bates correspondence]. 

16. See generally, S. Epstein, The Politics of Cancer (1979); EDF and Boyle, Malignant 
Neglect (1979). 

17. See Merrill and Hutt, Food and Drug Law: Cases and Materials 10-11, 439-452, and 
520-528, (1980). 

18. See text accompanying notes 139-144 infra. 

19. Merrill, Consumer Products at 1306-1309. 


identifying candidates for new health standards and scheduling regulation of 
those found to be carcinogenic.^^ 

These several avenues by which agencies identify chemicals that become 
targets for regulation are not mutually exclusive. An agency may have a role in 
deciding which chemicals are tested, and thus an opportunity to anticipate ex- 
perimental findings. The report of an industrial accident may converge with 
laboratory findings that press, and permit, an agency to set control limits, or it 
may precipitate the testing necessary to assess the new exposure intelligently. 
Furthermore, even an agency determined to devote its limited resources to the 
chemicals that it considers most important will be diverted by scientific discov- 
eries and public controversy.^^ 

Among the several sources of candidates for regulation, animal experiments 
have been the most important in recent years. Agency review of substances 
previously approved or otherwise potentially within their jurisdiction may even- 
tually provide more systematic information and thus enhance the rationality of 
regulation. It should not be expected, however, that society will soon develop 
fool-proof systems for detecting human cancer risks. Most of the chemicals to 
which humans are exposed are not clearly identified; many are of natural origin, 
the results of combustion, or unidentified contaminants of commercial products. 
We have not yet discovered how to incorporate these types of materials into our 
regulatory structure. ^^ 

The manner in which a chemical comes to official attention significantly 
affects the adequacy of the data available for evaluating its hazards. The four 
regulatory agencies do little safety testing of individual chemicals. ^^ They par- 
ticipate in the selection of chemicals tested under other federal auspices, '^'^ and 
they can influence private decisions about what to test.^^ But most testing of 
chemicals that are candidates for regulation is done outside government — by 
university scientists, commercial laboratories, or manufacturers themselves. 

If a manufacturer needs government approval before it can commercialize 
a chemical, the responsible agency is in a position to demand comprehensive 
testing. If, however, the agency must initiate the process for producing a decision 
about whether to limit exposure, generally it must rely on less complete data. 
To be sure, conscientious producers will perform some tests of their own. TSCA's 
premarket notification requirement^^ may elicit more testing, as well as reporting 

20. OSHA Draft at 87; text accompanying notes 119-130 infra. 

21. See Merrill, Consumer Products at 1323-1332. 

22. Bates Correspondence at 9. 

23. Only FDA and EPA have authority to require premarket testing, and these powers are 
limited to certain categories of products. While both agencies have substantial scientific capabilities, 
they have understandably focused their attention on test methods, rather than specific substances. 

24. National Toxicology Program, NTP Technical Bulletin 4: 3-6 (1981). 

25. Under TSCA, EPA can mandate testing of chemicals. Other programs provide only the 
opportunity to lobby for industry testing of problem chemicals. 

26. See 15 U.S.C. § 2604. 


of tests that are conducted, for chemicals that do not require premarket approval. ^^ 
But these mechanisms are unpredictable and unlikely to produce comprehensive 
data for individual chemicals or uniform data for any class. It should not be 
surprising, therefore, that information about the health effects of chemicals that 
have been regulated has invariably been incomplete and often outdated. -^^ 

This problem is accentuated when an agency is called on to evaluate a 
chemical on the heels of an industrial accident or a new laboratory finding. The 
agency usually has had no opportunity to anticipate what questions it might need 
answers to, much less commence gathering information. In some cases it may 
not even know what substance it is dealing with. For example, when illness was 
first reported among Michigan dairy cattle, the agent responsible was unknown. 
Once federal and state investigators had determined that PBBs had somehow 
got into cattle feed, scientists turned to assess the chemical's capacity to cause 
harm — only to discover a vacuum of information.^^ FDA's initial estimate of 
the health hazard posed by PBB exposure was therefore based on extrapolation 
from then-available data on a related compound, polychlorinated biphenyls 
(PCBs).^^ And its regulatory response was dictated solely by its capacity to detect 
the chemical.^' It took several more months to conduct studies of Michigan 
residents opposed to PBBs, and still longer to complete animal experiments. 

Although an unusual case, the PBB episode dramatizes the problems that 
sometimes confront an agency when a chemical first becomes a candidate for 
regulation. The adequacy of scientific data for informed risk assessment will 
vary with the circumstances that focus an agency's attention on a chemical. It 
is unrealistic to expect that the process for regulating chemical carcinogens will 
escape these limitations. 

2. Determining Carcinogenicity of Candidates for Regulation. 

After identifying a chemical that is to be considered for regulation, an 
agency's attention shifts to determining the hazard it poses. In the present context 
this inquiry can be summarized m the general question: Is a chemical capable 
of causing cancer in humans? This question embraces several subordinate issues 
ranging from issues of pure fact (e.g.. How many studies of the chemical have 
been reported? What are their characteristics?), through issues of scientific judg- 
ment (e.g.. Were the doses administered too high? Are observed lesions can- 
cerous?), to mixed issues of science and policy (e.g., Should benign and malignant 
tumors be aggregated?). 

27. Only food additives, color additives, human and animal drugs, and pesticides currently 
require premarket approval. 

28. Hutt. Safety Regulation in the Real World, 28 F.D.C.L.J. 460 (1973) [hereafter Regulation 
in the Real World]. 

29. Merrill and Schewel, Food Contaminants at 1412-1414. 

30. Id. at 1411. 

31. Id. at 1412. 


Determining whether a chemical is capable of causing human cancer in- 
volves three distinct stages which, in the absence of an emergency, may consume 
many months or even years. First, the agency must determine what evidence 
concerning the substance's carcinogenicity is available. Second, it may attempt 
to generate additional evidence, e.g., by requiring distributors of the chemical 
to conduct tests, by requesting that another government body conduct tests, or 
even by conducting tests itself. Finally, having decided that the available evidence 
is adequate to permit a judgment or that, in any case, the evidence cannot be 
augmented, the agency must decide whether the chemical can cause cancer in 

These analytically distinct steps are often merged in practice. The events 
that bring an agency's attention to a chemical may include the discovery that 
exposure is associated with a sharp increase in human cancer. Or test results 
showing a high probability of carcinogenicity may initially trigger regulatory 
concern. Moreover, an agency sometimes must decide quickly whether a chem- 
ical poses a serious health risk, without any practical opportunity to seek ad- 
ditional evidence. The world of regulation thus differs sharply from the deliberative 
model. •'^ 

It is nonetheless useful to keep in mind the distinction between assessment 
of the adequacy of evidence to permit a judgment about carcinogenicity and 
interpretation of evidence that is regarded as adequate to permit judgment. Sci- 
entific expertise and objectivity are essential in both functions. Deciding whether 
evidence is adequate requires understanding of the types of studies done to 
determine carcinogenicity and the technical competence to apply that under- 
standing in evaluating particular studies. But deciding whether the agency should 
delay a decision for better evidence also requires political sensitivity and aware- 
ness of budgetary and legal limitations.^^ 

Once it is determined that a decision about carcinogenicity must be reached, 
the issues to be resolved require primarily scientific judgment. Animal experi- 
ments and, when available, human epidemiological studies must be evaluated 
for adequacy of design, execution, and interpretation of results. Most charcter- 
istics of such evidence are objectively verifiable. It should not be difficult, for 
example, to determine how many animals were used in an experiment or how 
many were subjected to pathological examination. Some characteristics may as 
a practical matter elude verification because evidence about what happened has 
disappeared, e.g., did animals in the high dose group receive the high dose diet 

32. See Hutt, supra, at 467-68. 

33. There is often ambiguity in reports of toxicological studies and always some finite pos- 
sibility, even if small, that the experimental data, even if unambiguous, may be the result of chance. 
More testing may strengthen or weaken the evidence. Scientists are well-equipped to describe the 
strength of the evidence, but not obviously better qualified to determine how strong the evidence 
should be to justify regulation. This is properly a matter for policymakers, e.g., Congress or the 
agency. Congress may, for example, indicate a preference in favor of action to control exposure to 
a substance whose carcinogenicity for scientists remains in doubt. 


everyday? But the basic parameters of pertinent studies can usually be determined 
without controversy. 

Interpreting studies whose objective parameters are agreed upon poses more 
difficult problems. Whether the dose levels administered in an experiment were 
appropriate is a matter of judgment. Whether a lesion discerned in one organ of 
a single animal is neoplastic, i.e., is cancerous, is likewise a matter of judgment 
on which expert pathologists may disagree. Issues of this kind are ubiquitous in 
the interpretation of carcinogenicity tests, but they are no less scientific issues 
even though ultimately judgmental. 

A third kind of issue that arises in the assessment of carcinogenicity cannot 
be characterized as purely scientific. Suppose, for example, that experts are 
evenly divided about whether a study demonstrates a statistically significant 
increase in tumors among animals exposed to the test chemical. Is the study to 
be interpreted as confirming carcinogenicity? The issue of what degree of as- 
surance should be required that an apparent association is real is both technical — 
what is the statistical probability of a false result? — and political — what degree 
of assurance is required, as a matter of law or policy, to support a decision? 
Another example of this kind of mixed issue is posed by evidence that a chemical 
causes benign tumors. Toxicologists can provide experts opinions about the 
probable correlation between benign tumors and eventual malignancy, for a 
specific test or generally, but they have no unique claim to deference on the 
issue of whether benign tumors should, as a matter of policy , generally be equated 
with malignant tumors in interpreting test results. Resolution of this issue turns 
in part on considerations in which regulators are more expert — such as inti- 
mations of Congressional risk-averseness and the administrative costs of reliti- 
gating the issue for each chemical. 

This discussion reveals that the issues confronted in determining whether 
a chemical is capable of causing human cancer are not exclusively scientific. 
But this caveat should not obscure the fact that they are dominantly scientific. 
Accordingly, it will be urged later that the internal processes by which agencies 
make this determination initially and the procedures for resolving the issue 
publicly should feature participation by scientists and elicit conclusions which 
are couched in scientific terms. These objectives do not necessarily dictate re- 
moving evaluations of carcinogenicity from regulatory auspices; they do suggest 
insulation from policy formation and adversarial contest. -^"^ 

34. See NAS/NRC, Risk Assessment in the Federal Government: Managing the Process 131- 
140 (1983) [hereafter Risk Assessment in the Federal Government]. 

35. The issue can be broken down into several issues, some scientific, some policy: (1) Are 
the pathologic criteria good enough to distinguish benign from malignant tumors with a high degree 
of accuracy? This is a question for scientists. If the answer is no, an agency must decide, as a matter 
of policy, how to deal with the uncertainty. (2) What is the probability that a chemical causing only 
or mostly benign tumors in one experiment will do the same under all other circumstances? This is 
a question for scientists. (3) Should the risk of benign tumors be dealt with in the same way as the 
risk of malignant tumors, assuming that the risks can be distinguished experimentally? This is a 
question of policy. Comments of Richard Bates, M.D., Clement Associates and formerly FDA 
Associate Commissioner for Science, December 12, 1981. 


3. Assessing Exposure to Chemicals Targeted for Regulation. 

An essential step in deciding whether to regulate a chemical shown to be 
capable of causing cancer"^^ is to determine whether humans are exposed in a 
manner that creates cancer risk. Ordinarily this inquiry also entails some attempt 
to determine the characteristics of exposure, e.g., its route, magnitude, and 
duration. Measurement or estimation of these characteristics is essential if an 
agency wishes to quantify the risk posed by a carcinogen. ^^ 

Despite their obvious importance, however, issues of exposure have often 
received cursory treatment by regulators. FDA's estimates of human exposure 
to food contaminants have often rested on scanty residue data and assumptions 
about the average individual consumption of contaminated food.-^^ The CPSC 
has twice regulated asbestos in consumer products without extensive empirical 
investigation of human exposure.''^ EPA has devoted greater effort, in conjunc- 
tion with USDA, to estimating human exposure to pesticides that are candidates 
for cancellation. "^^ OSHA has made significant investments in determining ex- 
posure levels when its health standards had to be calibrated to different jobs. 

One explanation for the fragile character of many exposure estimates is that 
reliable data are not readily available from existing sources. In addition, they 
are rarely easy to obtain. Determining the characteristics of human exposures to 
chemicals draws on several technical skills, including chemical analysis, engi- 
neering, mathematics, and computer modeling. Ideally, actual measurements of 
human exposure should be made, but constraints of cost, time, and technology 
often prevent this. For example, FDA cannot practicably measure the amount 
of PCBs in each fish shipped in interstate commerce. Furthermore, any assay 
method for a substance has a lower limit of measurement that prevents reliable 
detection of smaller quantities."^" And patterns of use and exposure for some 
products, such as pesticides, may vary so widely that actual measurements at 
one location or of one population may not be representative of others. Accord- 
ingly, in addition to attempting to measure actual use and exposure patterns, 
agencies rely on mathematical models and similar tools to estimate the charac- 
teristics of exposure to carcinogens. 

The constraints on reliable estimation of exposure are practical rather than 
theoretical. It often simply is not feasible to conduct the kinds of investigations 
necessary to determine exposure patterns to a chemical precisely. But with 
additional expenditures and improved measurement techniques, it would be pos- 
sible to narrow the range of uncertainty considerably. Exposure estimates, unlike 
assessments of carcinogenicity, are based on exposure of the species that is of 
immediate interest, i.e., human beings. Disputes over carcinogenicity usually 

36. See text accompanying notes 321-328 supra. Part I. 

37. See Merrill & Schewel, Food Contaminants at 1396-97. 

38. See Merrill, Consumer Products at 1332-1339, 1346-1351. 

39. See text accompanying notes 310-321, supra. Part I. 

40. See text accompanying notes 322-325, supra. Part I. Merrill and Schewel, Food Con- 
taminants at 1397-1398. 


center on the implications of data that, though the best obtainable, are inherently 
deficient. Disputes over exposure tend to revolve around the adequacy of mea- 
surements that could be improved. 

Determining exposure can be characterized broadly as a technical exercise. 
Indeed, theoretically this element is the most susceptible of impartial resolution. 
While there ordinarily is room for debate about the accuracy of measurements 
or the adequacy of models, at bottom disagreement is the product of technical 
judgment rather than controversy over social policy. Experts may differ in their 
predictions of likely human behavior, e.g., whether pesticide applicators follow 
label directions, but this kind of disagreement can be resolved through more 
careful study of actual behavior. Exposure estimates should incorporate a range 
of error, but they need not embody social judgments. 

4. Quantitative Assessment of Risk. 

Analytically this step follows characterization of human exposures to a 
chemical; to perform a quantitative assessment of risk, an agency requires both 
dose-response data for the chemical and information about the extent of human 
exposure to it. In practice, however, qualitative assessment of carcinogenicity 
and quantification of the risk posed by different exposures are often performed 
contemporaneously . ^ ^ 

Risk quantification is optional under some regulatory approaches. The CPSC 
once spumed quantitative risk assessment in regulating carcinogens in consumer 
products, though more recently it has found the technique useful in determining 
whether a risk is unreasonable."^^ OSHA has consistently declined to use quan- 
titative risk assessment in setting worker exposure limits for carcinogens."^^ EPA, 
on the other hand, has long used quantitative risk assessment to set agency 
priorities, evaluate the risk remaining after application of controls, and to aid 
comparison of risks and benefits under FIFRA."^ 

Though indeed complex and potentially controversial, quantitative assess- 
ment of risk is a data-using rather than fact-gathering exercise. Rather, it com- 
bines data on a chemical's health effects and evidence describing the characteristics 
of exposure to estimate possible cancer incidence under different exposure con- 
ditions. It is, in short, a technique for analyzing and interpreting data supplied 
by other disciplines, including toxicology, epidemiology, engineering, and math- 

Several elements of this exercise excite sharp disagreement. Some critics 
have objected to quantitative risk assessment on principle, questioning both the 

41 . EPA's Cancer Assessment Group, for example, evaluates the carcinogenicity of substances 
and prepares a quantitative assessment of risk for all parts of the agency. See Risk Assessment in 
the Federal Government at 105-109. 

42. Merrill: Consumer Products at 1365-1366; text accompanying notes 377-380, supra. 

43. OSHA Draft at 95-98; text accompanying notes 381-387, supra; Part I. 

44. Text accompanying note 374, supra. Part I. 


intellectual coherence of the formulas used and the legitimacy of any formula 
that facilitates assigning a value to human life."^^ But dispute does not end here. 
Quantitative risk assessment requires two extrapolations that generate disagree- 
ment on both technical and policy levels. 

First, an agency must choose a formula, or "scaling factor," for extrap- 
olating to man from data obtained in experiments using small rodents who live 
for perhaps 30 months. Four scaling factors enjoy some support among scientists. 
The choice of scaling factor depends primarily on scientific criteria, and might 
vary with the route of human exposure, "^^ but it can affect risk estimates by a 
factor of 40."^^ This disparity may invite demands that an agency select the most 
"conservative" scaling factor whenever scientific criteria do not dictate a choice. 

The choice of a model for extrapolating from high to low doses levels 
occasions similar debate. Though different biological rationales support the var- 
ious models, none has elicited consensus among either scientists or regulators. "^^ 
Closeness of fit with the experimental data rarely provides a basis for preferring 
one model. '^^ Because the risks predicted at low doses can diverge by several 
orders of magnitude, ^^ agencies are again pressured to select the most "con- 
servative" model, i.e., the one that is least likely to underestimate risk. 

Quantitative assessment of the risk of carcinogens raises numerous other 
issues of regulatory policy. It is difficult to disaggregate such issues from issues 
of technical adequacy and scientific judgment at this stage of analysis. So long 
as agencies leave such issues as the choice of scaling factor and extrapolation 
model to be resolved in proceedings to regulate individual chemicals it would 
be difficult to contend that the estimation of the magnitude of the risk posed by 
a specific substance should be left to scientists. ^^ 

5. Evaluating Costs of Control. 

This step in the analysis draws on different skills from those required to 
evaluate a chemical's risks. Engineering and economics are the primary disci- 

45. Text accompanying note 368, supra, Part I. 

46. Office of Technology Assessment, Assessment of Technologies for Determining Cancer 
Risks From the Environment at 169-170 (1981) [hereafter OTA Report]. 

47. Id. An agency may, of course, compute the estimated risk of a carcinogen using more 
than one scaling factor, thus generating a range of possible risks. So doing, however, invites demands 
that it adopt the formula that yields the highest risk for that substance, unless there are scientific 
grounds for preferring another. If such grounds exist, it is unclear why the agency would not select 
that scaling factor in the first instance. The explanation is that there rarely is a convincing scientific 
basis for adopting one scaling factor and rejecting others. 

48. Currently the linearized multi-stage model seems to have achieved primacy among the 
regulatory agencies. 

49. OTA Report at 162; text accompanying note 353, supra. Part I. 

50. E.g., OTA Report at 163; OSHA, Identification, Classification and Regulation of Potential 
Occupational Carcinogens, 45 Fed. Reg. 5001, 5184-95 (1980) [hereafter Carcinogen Policy]. 

51. OSHA's carcinogen policy was an attempt to resolve these issues once and for all future 
standard-setting proceedings. Theoretically, adoption of the agency approach would narrow the 
issues in such proceedings to ones that were primarily scientific. See OSHA Draft at 108-109. 


plines involved, but others are also useful because the "costs" of controlling 
human exposure to chemicals take many forms. Familiarity with technologies 
for reducing exposure is important in such programs as OSHA's, where direct 
expenditures for exposure controls — in capital investment and annual mainte- 
nance — comprise the primary costs of regulation. For FDA, the main measure 
of the costs of reducing human exposure to food contaminants is the value of 
above-tolerance food that cannot be marketed. ^^ For the CPSC, the costs range 
from the value of inventory that can no longer be sold,^^ to loss of future sales, ^"^ 
to the increased expense of substitute ingredients,^^ to the lost enjoyment of 
products for which there is no substituted^ 

The potential scope of what I have termed "control costs" has often prompted 
agencies either to narrow the effects they will consider or to perform the inquiry 
in perfunctory fashion. For example, FDA has announced that it will not consider 
the "secondary effects" of tolerances for food contaminants, e.g., the indirect 
but possibly substantial impact on businesses involved in recreational fishing. ^^ 
The CPSC's formal assessment of the economic consequences of banning TRIS- 
treated garments was cursory, ^^ and it gave little attention to the costs of recalling 
hairdryers insulated with asbestos. ^^ On the other hand, OSHA and EPA have 
both paid increasing attention to the costs of occupational health standards and 
alternative restrictions on pesticide use. EPA has created a special group of 
agency economists to evaluate the economic effects of actions taken throughout 
the agency. ^^ 

The four agencies have been criticized for underestimating control costs. 
OSHA's estimate of the cost of compliance with its benzene standard — over 
$400 million — was approximately one-fifth of the industry's estimate.^' Simi- 
larly, its estimate of the costs of its standard for cotton dust was far lower than 
the industry's figure, and relied on an industry-financed study whose assumptions 
and methodology the industry disavowed. Estimates of control costs often vary 
over a wide range. Disagreement is unrestrained in part because the methods 
for estimating costs are not so well standardized as the experimental protocols 
for assessing health effects. ^^ Debate also rages over starting assumptions re- 
specting industry structure, suitability of substitute materials, and likely behavior 

52. Merrill and Schewel, Food Contaminants at 1392-1398. 

53. See, e.g., Merrill, Consumer Products at 1323-1332. 

54. See, e.g., 46 Fed. Reg. 11188, 11200 (1981). 

55. See, e.g., Merrill, Consumer Products at 1339-1346. 

56. See, e.g., id. at 1332-1339. 

57. Merrill and Schewel, Food Contaminants at 1418; 44 Fed. Reg. 38330 (1979). 

58. Merrill: Consumer Products at 1325-1327. 

59. Id. at 1349. 

60. See discussion of Elizabeth Anderson with NAS Committee on Institutional Means of 
Risk Assessment, October 13, 1981. 

61. 5£'^IndustrialUnionDept.,AFL-CIO V.American Petroleum Inst., 100 U.S. 2844(1980). 

62. Compare debate over the economic effects of OSHA's standard for vinyl chloride, OSHA 
Draft at 147-151. 


of firms under different control options, and the appropriate rate at which to 
discount future expenditures.^^ 

Perhaps the most serious flaw in estimates of control costs is author bias, 
which the lack of standardized methods of analysis facilitates. Agencies generally 
are interested in minimizing the economic impact of control measures; producers 
and distributors have even greater incentives to exaggerate these effects. This 
problem is aggravated by unequal access to necessary information. Firms that 
manufacture or use a chemical ordinarily are better equipped than regulators to 
evaluate substitutes or anticipate technological changes than can dramatically 
influence the costs of compliance. Moreover, they usually regard data concerning 
production and sales levels and internal costs as confidential. 

Estimating the costs of controlling exposure to a chemical requires sophis- 
ticated expertise, but it is also vulnerable to partisan evidence and argument. It 
therefore should be a product of procedures that permit participation by divergent 
interests. Furthermore, estimating costs is analytically separate from qualitative 
and quantitative assessment of risk. This distinction should be punctiliously 
observed in agency decisionmaking.^"^ 

6. Selecting Measures for Controlling Exposure. 

Of the several steps in my prototype analysis, this is the least technical. To 
be sure, agency decisions should reflect careful study of the character and mag- 
nitude of the risk posed by a chemical and of the economic costs of different 
levels of control. But in the absence of statutory directions to adopt only those 
controls whose health benefits outweigh economic costs, ^-^ determining what 
level of exposure is "acceptable" is at bottom a value judgment. ^^ Indeed, it is 
doubtful that a Congressional mandate to adopt only those controls whose benefit- 
cost ratio is greater than unity would alter the political character of the choices 
authorized by the programs studied. Methods for evaluating carcinogenicity 
incorporate judgmental elements, as to models for quantifying risk. The predic- 
tion of economic costs of control invariably is open to debate and susceptible 
to partisan manipulation. An experienced regulator can usually find numbers to 
support a range of decisions. 

If this characterization is roughly accurate, it has implications for the kind 
of process by which controls measures are chosen. Obviously the process must 
afford an opportunity for all affected interests to make arguments to the deci- 
sionmaker. Personal expertise is likely to be of only modest help; what the 
regulator needs is access to independent expert advice to inform judgments that 
may in the final analysis be intuitive. 

63. See Lave, The Strategy of Social Regulation 29, et seq. (1981). 

64. See generally Nichols and Zeckhauser at 55-57. 

65. No statute specifically directs such analysis. See text accompanying note 94, supra. Part I. 

66. See, e.g.,W. Lowrance, Of Acceptable Risk at 8 (1976). 


C. Setting Agency Priorities 

As the previous discussion suggests, agencies have chosen chemicals for 
regulation more often as the result of external events than independent investi- 
gation. With increasing awareness of the number of chemicals to which humans 
are exposed and the proportion that are capable of causing cancer, agencies have 
become aware of the need to proceed more systematically. This section starts 
from the assumption that criteria for systematic priority setting can be devised. 
It also assumes, however, that regulators must also be able to respond to un- 
solicited information that calls for investigation and possibly for prompt regu- 
latory action. Not only does the public health require flexibility to deal with 
unanticipated hazards; regulators are subject to political pressures that often will 
ignore any planned agenda. 

/ . Need for Priorities . 

Government efforts to protect the public from chemical hazards include 
both information gathering (including testing) and exposure control. Both activ- 
ities require systematic planning, because the potential workload is considerably 
larger than the resources available to perform either. ^^ 

The federal government funds many programs that are designed to obtain 
data about the toxicity of individual chemicals. These data are valuable for 
regulatory decisionmaking, though they have other uses as well. Basic research 
in chemistry and the biological sciences aides regulators, but little of it is con- 
ducted with regulation in mind — nor should it be. The concern here is with those 
information gathering activities that are performed or required by the government 
to assist regulatory decisionmaking. One example is the bioassay program of 
the National Toxicology Program (NTP). Established in 1978 by then-HEW 
Secretary Califano, NTP is a coordinating umbrella for toxicological research 
carried out by diverse components of the Department of Health and Human 
Services and other government bodies. ^^ NTP has assumed responsibility for 
overseeing most of the long-term and much of the medium-term health effects 
testing of chemicals performed by the federal governments^ — including the bioassay 
program previously administered by the National Cancer Institute. This program 
has become the primary source of federally funded-data about the chronic toxicity 
of chemicals. 

In addition to funding health effects testing, the federal government elicits 
many of the tests performed by private industry. Because they administer pro- 
grams requiring premarket approval of new chemicals (e.g., drugs, food addi- 

67. See, e.g.. National Research Council, Steering Committee on Identification of Toxic and 
Potentially Toxic Chemicals for Consideration by the National Toxicology Program, Strategies to 
Determine Needs and Priorities for Toxicity Testing: Design (1981) (hereafter referred to as NTP 

68. Id. at 1, 5. 

69. Id. at 5-9. 


tives, and pesticides), FDA and EPA are able to demand considerable testing 
by industry. This authority is not confined to new compounds. EPA has explicit 
power to require new tests of already registered pesticides^^ and FDA is often 
able to elicit private testing of old compounds.^' Furthermore, under TSCA, 
EPA also has authority to mandate toxicity testing for chemicals in development 
or already in distribution, even though they are not subject to premarket ap- 
proval. ^^ The large number of such chemicals makes systematic priority setting 
for this latter activity crucial. ^"^ 

The other activity that requires thoughtful planning is the selection of chem- 
icals for evaluation and regulation, e.g., for eventual establishment of new 
exposure limits. Until recently, most of the chemicals regulated by the agencies 
studied here were first identified by external events, rather than discovered 
through systematic planning. 

2 . Purposes for Setting Priorities 

Resource limitations are the main reason that agencies require criteria for 
selecting chemicals for testing and regulation. Society lacks knowledge about 
the health effects of most chemicals to which humans are exposed. There are 
exceptions, of course; the health effecis of agents used to treat disease are 
comparatively well understood. ^"^ Ignorance about the effects of other chemicals 
is vast, however, and information about human exposure is comparably sparse. 
For example, we have only rough estimates of the number of man-made chemicals 
humans are exposed to. The inventory EPA compiled following enactment of 
the Toxic Substances Control Act includes over 60,000 different compounds, 

70. FIFRA § 3(c)(2)(B), 7 U.S.C. § 136a(c)(2)(B) (1976). 

71. FDA has tentatively approved additional uses for a particular compound but on the 
condition that further testing be conducted {e.g. , L-Dopa). The agency also has threatened to terminate 
the provisional listing of certain color additives absent additional private tests. Merrill and Hutt at 

72. TSCA § 4. 15 U.S.C. § 2603 (1976). 

73. In compiling the inventory required by TSCA. EPA has catalogued more than 55,000 
substances that are produced in this country or imported and that are used in various phases of 
chemical manufacture and formulation in the United States. Section 4 of TSCA created the Inter- 
agency Testing Committee (ITC) to aid the Administrator of EPA in identifying chemicals already 
in commerce that should undergo testing for health and environmental effects. The ITC list of 
chemicals to be tested cannot exceed 50 at any one time, while the ITC updates its roster of chemicals 
every six months. NTP Report at 5, 105. 

"Although TSCA stipulates that the ITC shall rank the chemicals that it recommends for 
testing, the ITC has chosen not to do so, on the grounds that all designated chemicals are 
to be of equal priority for testing. The ITC has, however, developed a priority-setting 
process by which chemicals are initially scored by experts for exposure potential. High- 
scoring chemicals are then scored for health-effects potential, and the chemicals scoring 
highest at that stage are scrutinized individually for final selection. 

74. The Federal Food, Drug, and Cosmetic Act [hereafter FD&C Act] § 505, 21 U.S.C. §355 
(1976), requires premarket demonstration of safety and efficacy of new drugs, which has meant 
extensive testing of these agents in both animals and humans. See Merrill and Hutt at 404-09. 


but may yet be an underestimate. ^^^ There is little reason to think that many of 
these chemicals will be hazardous in ordinary use, but determining which ones 
should be more thoroughly tested and evaluated is no easy task. 

Social resources available to test and evaluate chemicals are relatively mea- 
ger. The cost of single-species bioassay for a chemical now exceeds $500,000. 
Furthermore, the facilities do not exist to test large numbers of chemicals, and 
toxicologists and pathologists equipped to design and evaluate animal tests are 
in short supply. 

Resource constraints also limit the selection of chemicals for regulation. 
No agency has been able to regulate more than two or three controversial chem- 
icals in any year.^^ Regulatory proceedings devour the time of agency scientists, 
lawyers, and economists at a rate that increases as the process becomes more 
formal and adversarial.^^ If agencies can regulate only a few chemicals, they 
must select their targets wisely. Systematic planning may help refute their rep- 
utations for unpredictability. In addition, thoughtful selection of candidates for 
regulation may facilitate coordination among agencies and yield administrative 

Rational priority setting would seem equally desirable from the vantage of 
firms whose activities are targets for regulation. The ranking of candidates for 
regulation should assist planning of production and equipment repair and re- 
placement. The advantages of systematic priority setting have sometimes been 
questioned by industry spokesmen. Individual firms and trade groups have vig- 
orously criticized agency proposals to develop criteria for selecting, and an- 
nouncing, targets for regulation. ^^ Their opposition may betray a natural skepticism 
of any initiative that may make regulation easier. However, many firms also 
voice separate concern about the commercial impact of public announcements 
of agency priorities. Pointing out that an agency's announced plan to regulate 
chemical X does not resolve the issues whether the chemical poses a health 
hazard or what exposure limits are appropriate, producers contend that the public 

75. NTP Report, at 5. Chemical Manufacturers Association, The First Four Years of the Toxic 
Substances Control Act: A Review of the Environmental Protection Agency' s Progress in Imple- 
menting TSCA 37-45 (1981). 

76. OSHA, for example, has issued fewer than twenty ocupational health standards since its 
inception. See Risk Assessment in the Federal Government at 97. The chlorobenzilate RPAR pro- 
ceeding took EPA nearly three years to complete, while the toxaphene RPAR was begun in 1977 
and has not yet been concluded. CPSC spent nearly four years developing its "final" regulation 
declaring vinyl chloride a banned hazardous substance when used in self-pressurized household 
products. Merrill, Consumer Products at 1311-18. 

77. See OSHA Draft at 22-26. 

78. Comments by industry spokesmen and trade groups are included and discussed throughout 
the Federal Register document containing OSHA's final version of its generic carcinogen policy. 
45 Fed. Reg. 5002 (1980). See, e.g., 45 Fed. Reg. 5036 ff., 5067 ff., 5079 ff., 5084 ff., 5094 ff., 
5097 ff., 5100 ff., 5105 ff., 5109 ff., 5115 ff., 5118 ff., 5139 ff., 5149 ff., 5153 ff., 5160 ff., 
5174 ff., 5179 ff., 5209 ff. It was the CPSC's effort to implement its system for evaluating and 
classifying chemicals that precipitated the successful court challenge to its own career policy. See 
Merrill, Consumer Products at 1300-04. 


identification of candidates for regulation can have devastating commercial im- 
pact before there is any opportunity to address the merits. ^^ This concern must 
be weighed in designing any program for identifying candidates for regulation. 
Another rationale has been offered for systematic priority setting. It has 
been suggested that human health would be better protected if agencies regulated 
more chemicals less stringently.^^ The premises of this theory are three: First, 
it is argued that the risk posed by most chemicals is likely to decline more than 
linearly with reduced exposure. Second, it is contended, toxic chemicals are 
likely to combine to produce health effects that are more than additive. Broad- 
ening the scope of regulation is a response to the hazards of chemical interaction. 
Third, less stringent controls are less likely to generate formal challenges at the 
administrative level and few court proceedings. In consequence, regulations 
should be able to deal with more potential hazards to health.*^' 

3. Criteria for Setting Priorities 

Considerable thought has been given in recent years to the criteria agencies 
should employ in selecting chemicals for testing and regulation. ^^ This section 
does not attempt to prescribe precisely the criteria that decisionmakers should 
follow; it does suggest some of the considerations they should weigh and de- 
scribes some promising experiments in priority setting. 

a. Selecting chemicals for testing. The same criteria may not be appro- 
priate for identifying chemicals to be tested and selecting candidates for regu- 
lation. In the first context, the primary objective is to reduce ignorance about a 
chemical's effects; in the second, the objective is to make efficient use of public 
and private resources for controlling exposure to chemicals found capable of 
causing harm. This distinction should not, however, obscure the fact that the 
testing of interest is done primarily to aid regulation. 

79. For example, the Society of the Plastics Industry. Inc. made the following post-hearing 
comment: "First of all, unevaluated lists of substances being considered as potential candidates for 
regulation as carcinogens should not be treated lightly. While the parties to OSHA's proposed cancer 
policy proceeding were ahxious to see the preliminary categorization, the publication of a list which 
had not been at all scrutinized is regarded by industry as inappropriate. . . . For example, the 
erroneous listing of a substance as being under consideration of regulation as a carcinogen can cause 
the two extremes of public reaction-one, concern about the use of the material and, the other, 
erosion of the credibility of governmental and other scientists in assessing carcinogenic hazards. In 
addition the scientific resources of government, industry, and the public are diverted to investigating 
and rebutting erroneous classifications. Furthermore, the incorrect association of materials, which 
have been shown to be innocuous and safe for their intended uses, with well known, potent carcin- 
ogens can have a detrimental effect on their manufacture, marketing and use. For such reasons, 
OSHA should take much greater care in this area." 45 Fed. Reg. at 5209. 

80. J. Mendeloff, Regulation of Toxic Substances: Is Less Better? (1981) (unpublished draft 

81. Id. 

82. See NTP Report at 1 1-34. 5^^ e.g., Weinstein, Decision Making for Toxic Substances 
Control: Cost-Effective Information Development for the Control of Environmental Carcinogens, 
27 Public Policy 333 (1979). 


A hypothetical example reveals the close relationship between the two 
functions. Suppose that an agency confronts one hundred chemicals whose tox- 
icity is unknown and which it could order to be tested. If one had reason to 
suspect that as many as 15 would be found carcinogenic in animal bioassays, it 
would be desirable to maximize the likelihood of finding those 15 early in the 
sequence of testing. In such circumstances a negative result would provide less 
useful knowledge than a positive result. Accordingly, the criteria used in selecting 
chemicals for testing should enhance the likelihood of identifying the 15 probable 
carcinogens sooner rather than later, so that regulators could employ the results 
in making regulatory decisions. This may require coordination between those 
responsible for selecting which chemicals to test and those responsible for reg- 

There are, however, numerous obstacles to effective coordination. First, 
regulatory agencies have been slow to develop criteria for identifying candidates 
for regulation. Any body responsible for choosing chemicals for testing would 
have difficulty discerning what chemicals agencies were interested in without 
being told. More important, the universe of chemicals that require testing is not 
only large but largely uncharacterized; it is not known how much information 
exists about the chemicals in the universe. 

This latter problem has bedevilled the government's major testing effort, 
the National Toxicology Program, which desires to develop a system for selecting 
chemicals for testing. ^"^ Before deciding what criteria should govern this selection 
the NTP has considered it essential to learn more about the toxicological data 
base for the entire universe of identified commercial chemicals. That universe 
consists of approximately 75,000 compounds: some 60,000 from the TSCA 
inventory plus approved pesticides, approved food additives and GRAS food 
ingredients, known cosmetic ingredients, and known and approved ingredients 
of drugs. ^"^ Much more information is likely to be available about chemicals in 
the latter groups than about chemicals in the TSCA inventory. At NTP's request, 
the National Academy of Sciences has undertaken a study to define representative 
examples of each segment of the universe, evaluate the adequacy of the toxi- 
cological data for each segment, and ultimately recommend criteria for selecting 
chemicals for further testing.^'' 

b. Selecting chemicals for regulation. Selecting candidates for regulation 
may be a smaller task, for fewer chemicals are likely to be involved, but is a 
more subtle one. Agencies have been urged to regulate the "worst first," but 
this general rubric is difficult to implement. The premise is that a rational 
regulator would proceed first against the most toxic materials. But the most toxic 
chemicals are not necessarily the "worst"; the risk posed by carcinogen depends 
not only on its potency but also on the number of people exposed to it and the 
manner, magnitude, and duration of their exposure. 

83. See NTP Report at 5, \\?>. 

84. /^. at 1,5. 

85. Id. passim. 


Before outlining the types of criteria agencies might use in selecting can- 
didates for regulation, some preliminary observations are in order. First, the 
same criteria may not be appropriate for all four agencies. While the programs 
studied do not illustrate the full range of federal efforts to regulate toxic sub- 
stances, they reflect important differences. The most significant is the most 
obvious; the four agencies have largely distinct jurisdictions. An agency's criteria 
must fit its jurisdiction. The jurisdictions of the four agencies are defined by 
medium of exposure (food or consumer products); by location (the workplace); 
or by chemical use (pesticides). One agency's jurisdiction may be open-ended; 
and any chemical that may contaminate food is of concern to FDA. Another's 
may be self-defining; all pesticides distributed in the United States must be 
registered with EPA. A further distinction is likely to be influential; that between 
chemicals that are already in the environment or approved for use, and chemicals 
yet to be developed or awaiting approval. All of the programs studied here have 
been concerned primarily with chemicals in the former category. 

While agency jurisdictions generally do not overlap, many chemicals have 
uses that bring them within the authority of two or more agencies. This fact has 
stimulated suggestions that priority setting be done on an interagency basis. ^^ 
An alternative is for each agency to take account of what other agencies are 
doing in its own setting of priorities. Finally, it should be stressed that all four 
agencies are concerned about adverse health effects in addition to cancer. Thus, 
any priority-setting system will inevitably incorporate some effort to evaluate a 
chemical's capability of causing such effects, perhaps sometimes subordinating 
potential carcinogenicity as a criterion for concern. 

The frameworks suggested for selecting chemicals as candidates for regu- 
lation share several common elements. Broadly speaking, they stress three cri- 
teria: (1) toxicity, i.e., potential for causing harmful effects; (2) exposure, including 
number of people exposed, the severity of their exposure, and its duration; and 
(3) what have broadly been described by one source as "socio-economic con- 
siderations," a category that embraces such concerns as availability of substi- 
tutes, the cost of control measures, and administrative feasibility. The first two 
of these address what usually is an agency's central concern: the health risk 
posed by a chemical. 

The health risk posed by a chemical is a function of its carcinogenic potency 
and of the characteristics of human exposure to it. Precisely because risk is a 
function of these two criteria, there are two possible starting places for evaluation. 
EPA, which must approve all pesticides before they may be distributed, can 
readily identify all such compounds subject to its control. Manufacturers of 
consumer products, however, need not obtain approval or disclose composition 
before distribution. For the CPSC to confine its search to lists of chemicals would 
be foolish. The agency has wisely begun to review consumer products to de- 
termine which might pose high risks //they contained toxic materials. 

86. See IRLG, Notice oflRLG Work Plans and Public Meetings, 43 Fed. Reg. 7174 (1978). 


Theoretically evaluation of exposures would be an important step in priority 
setting even if an agency knows which chemicals fall within its jurisdiction. A 
toxic material may present a very low risk if only a few people are exposed at 
low levels for brief periods of time. Because information about exposure to most 
chemicals is generally poor, however, this criterion is sometimes neglected. 

In addition to the criteria of toxicity and exposure, an agency would be 
wise to take into account the probable costs of controlling exposure to a chemical. 
Those costs could be as diverse as those the agency would explore more carefully 
when it commenced formal regulatory proceedings. This is not to suggest that 
agencies should refrain from regulating hazardous substances when to do so 
would impose high costs. But if resistance to regulation is likely to be propor- 
tionate to such costs, prudent use of agency resources justifies considering this 
factor. ^^ Another similar criterion is the likely administrative cost of setting and 
enforcing exposure limits. Agencies should not shrink from contested proceed- 
ings when health risks are high, but they should consider their ability to enforce 
any formal limit they set. It would be imprudent, for example, for FDA to initiate 
tolerance-setting for a food contaminant that it could not detect using available 
analytical techniques. 

Any priority system should permit an agency to add new candidates promptly. 
While petitions from interest groups or competitors are not perfect barometers 
of a chemical's importance, they are not obviously unreliable. Accidents may 
disclose graver hazards than an agency's advance planning, or political pressures 
may force a change in priorities. ^^ No system should impede response in these 
circumstances. Having a formal system for selecting candidates for regulation 
can, however, discourage demands to regulate low risk chemicals. 

The criteria suggested — toxicity, exposure, costs and administrability — 
should also be appropriate for priority setting on an interagency basis. If reg- 
ulation should address those chemicals that pose the greatest risk, rational priority 
setting should aggregate all of the settings in which exposure can occur. Other 
considerations are, however, likely to intrude when agencies with different con- 
stituencies collaborate. Jurisdictional jealousies and divergent experience may 
produce disagreement over what is important. For example, one might expect 
FDA and USDA to have difficulty agreeing about which food ingredients to 
control first. ^^ Worker awareness of the chemicals around which they work may 
prevent OSHA from ceding priority to an air pollutant to which a larger population 
is exposed. Furthermore, joint adherence to the suggested criteria could leave 
some agencies with only low-ranking candidates. Accordingly, when agencies 
collaborate in selecting chemicals for regulation, the final list is likely to include 
each agency's first choice. 

87. See Merrill, Regulation of Toxic Chemicals (Book Review), 58 Tex. L. Rev. 474-75, 
476, n.57. 

88. Substances brought to the forefront of regulatory consideration by accident or political 
pressure include PBBs, DES, 2,4, 5-T, benzene, Red No. 2, TRIS, and asbestos in hair dryers. 

89. USDA's principal concern typically would be to increase agricultural production while 
FDA's emphasis would focus more on public health protection. 


4. Agency Performance in Setting Priorities 

Separate chapters describe how the four agencies identified the chemicals 
they have attempted to regulate, and reiteration of those descriptions is not 
necessary here. Some discussion of the agencies' experience and their efforts to 
establish procedures for priority setting, however, will assist evaluation of per- 
tinent recommendations. 

a. Food and Drug Administration. The author has studied only one of 
several programs under which FDA regulates carcinogens: its setting of per- 
missible exposure levels for environmental contaminants of food.^^ Ordinarily 
these contaminants would not otherwise fall within FDA's jurisdiction; they are 
not subject to FDA premarket approval or used for other FDA-regulated purposes. 
FDA has an interest in their health effects only because they get into food, 
unintentionally and usually accidentally. Accordingly, the agency's method of 
selecting candidates for tolerances or action levels is calibrated to the circum- 
stances that give rise to its concern. 

Candidates for FDA regulation are a product of the agency's efforts to 
identify foreign substances in food and then seek information about their potential 
health effects. The latter function has typically been performed by FDA's sci- 
entific staff. The agency has never adopted a public system for identifying 
potential contaminants or eliciting information about their health effects before 
deciding whether regulation is appropriate. Rather, it has informally explored 
the scientific literature or consulted with individual scientists to assemble the 
information necessary to determine what level of regulation is appropriate.^' 

Because FDA has relied almost exclusively on informal action levels rather 
than formal tolerances, theoretically it has been able to "regulate" every con- 
taminant that it has identified, i.e., set a level at which it will seize contaminated 
food as adulterated. The action level for a chemical can be set at whatever level 
FDA is prepared to say "may render" the food "injurious to health. "^^ Because 
its enforcement actions are rarely contested,^'' the agency may even set this level 
at zero for a contaminant whose health effects are unknown or only poorly 
documented. The likelihood that it will ever have to support its charge of adul- 
teration is thus small. 

The system FDA has devised for assuring the safety of food features regular 
monitoring of food shipments by inspectors in the field. This system practically 
requires the agency to have some action level (or tolerance) for every chemical 
likely to be discovered in food. It cannot respond to federal or state inspectors, 
who discover contaminated samples, that it has set no level for a particular 
contaminant because the substance ranks low on its priority list. It must have 
an answer to the question: Can the food lawfully be distributed? 

90. See generally, Merrill and Schewel: Environmental Contaminants. 

91. Office of Technology Assessment, Environmental Contaminants in Food 37-38 (1979) 
[hereafter OTA: Food Contaminants]. 

92. 21 U.S.C. § 342(a)(1). See Merrilland Schewel, Environmental Food Contaminants. 

93. Merrill and Hutt at 665, 735. 


Accordingly, the "priority setting" in which FDA engages consists of 
deciding which contaminants to search for. Historically, FDA has discovered 
contaminants through three procedures: (a) routine surveillance sampling of food 
shipments; (b) special monitoring surveys of particular foods or regions or of 
specific contaminants; and (c) total diet studies of contaminants found in the 
average diet, as estimated by so-called "market basket surveys" conducted by 
USDA.^"^ These three mechanisms yield information about not only which chem- 
icals contaminate food but which foods are affected and at what levels. 

FDA's criteria for monitoring have recently been criticized by both the 
Office of Technology Assessment and an intraagency study group. ^^ The OTA 
and FDA's own work group faulted the agency's approach on two main grounds. 
First, they argued that FDA's routine monitoring programs were too reliant on, 
and thus circumscribed by, currently available methods of chemical analysis. 
By gearing its search efforts to existing measurement techniques, according to 
OTA, FDA was essentially ignoring other criteria such as the potential of chem- 
icals to contaminate food and their toxicity, which are more germane to its main 
objective of preventing distribution of harmful food.^^ The OTA report rec- 
ommended that the agency should weight its monitoring activities in favor of 
health risk and exposure potential, and adjust its efforts to develop or improve 
analytic methods to these criteria. ^^ FDA's own study group also recommended 
that its total diet study program should be refined to yield information about 
differential exposures based on age, gender, and special dietary habits. ^^ 

b. Consumer Product Safety Commission. Selecting candidates for reg- 
ulation has been one of the CPSC's major problems. ^^ None of the chemicals 
that the Commission has attempted to regulate was identified through systematic 
review of products within its jurisdiction. Each one was the subject of a public 
petition submitted before the agency acted, though some officials insist that 
most, if not all, had previously been targeted by the agency's own staff. ^^ 
Congress' recent repeal of section 10 of the CPS Act,'^^ which obligated the 
agency to respond to petitions within 120 days, should eliminate whatever di- 
version this provision caused. 

The CPSC's difficulties in deciding what hazards to regulate have not been 
attributable solely to the petition process, however, nor have they all been within 
its control. The CPSC's jurisdiction is enormous and the products potentially 
subject to regulation present an extraordinary variety of human exposures. The 

94. Study Group on FDA Residue Programs, FDA Monitoring Programs for Pesticide and 
Industrial Chemical Residues in Food 26-30 (1979) [hereafter FDA Monitoring Programs]. 

95. OTA: Food Contaminants at 5, 112. 

96. Id. at 1 12- 113; FDA Monitoring Programs at 26-28. 

97. OTA: Food Contaminants at 41-42. 

98. FDA Monitoring Programs at 23. 

99. See Merrill, Consumer Products at 1361-65. 

100. Id. at 1363. 

101. Id. Sit 1314. 


agency lacks an inventory of consumer products, a roster of firms that manu- 
facture and distribute them, and any easy mechanism for forcing disclosure of 
their composition. 

Recognizing these difficulties, the CPSC in 1979 established an internal 
working group to identify and evaluate chronic health hazards for possible reg- 
ulation. '^^ The screening group originally concentrated on lists of potential hu- 
man carcinogens compiled by other bodies, e.g., the list of human carcinogens 
assembled by the International Agency for Research on Cancer (lARC) and the 
list of chemicals found positive in NCI bioassays. '^^ For each chemical the group 
planned to resolve, sequentially, a series of legal and scientific issues: (1) agency 
jurisdiction, (2) carcinogenicity, (3) extent of use in consumer products, (4) 
probable human exposures, and (5) regulatory options. It is not clear what criteria 
the group used for selecting from among several hundred listed chemicals those 
to be subjected to this rather costly review process. The group's plan did not 
call for Commissioner-level decisions about the priority of specific chemicals 
until after it has explored the first four of these issues. Under this arrangement, 
the Commission's staff could invest substantial effort investigating the effects 
and prevalence of a chemical before learning where it ranked in the Commis- 
sioners' larger set of priorities. '^"^ 

In 1979 the Commissioners voted to implement a recommendation from 
EDF that the chronic hazards screening group should broaden its focus and look 
at human exposures to consumer products as well as at chemicals identified as 
potential carcinogens. ^^^ Since the risk posed by a chemical is as dependent on 
the route, duration, and magnitude of exposure as on its potency, EDF had 
argued, the CPSC should also search for products to which consumer exposure 
could present a serious risk if they contained carcinogens. 

The CPSC's agenda for fiscal 1982 includes several chronic health hazards. 
These include UF foam insulation which the agency has recently voted to ban,'^^ 
and asbestos, into whose use the agency began a comprehensive investigation 
in 1979. •'^^ It is not clear whether either of these candidates for regulation was 
identified through the agency's own internal review process. Asbestos had twice 
previously been a target of Commission action, and both it and formaldehyde 
were selected for interagency attention by the IRLG.'^^ Several substances stud- 
ied by the screening group had been advanced for review by the Commissioners, 
only later to recede from their priority list. 

102. Id. at 1364-65. 

103. Id. at 1307. 

104. The chronic hazard program's priorities have always competed for Commission attention 
with several other types of hazards, even the Commissioners placed them in their annual program 
plan for special attention. 

105. Id. at 1307-08. 

106. Id. at 1356-60; [1981] Chemical Regulation Reporter (BNA) 33 (1982). 

107. Merrill, Consumer Products at 1351-54. 

108. IRLG Reports 2(1): 1-5, 17-19 (1980). 


It is appropriate to ask whether the CPSC should continue independent 
efforts to identify chemicals for regulation. It might be a more prudent use of 
its shrinking resources for the agency to concentrate on chemicals considered 
for regulation by other agencies with larger scientific staffs and devote its own 
efforts to determining exposure to, and the risk of, consumer products in which 
those chemicals occur. The problem with this approach is illustrated by for- 
maldehyde. If sister agencies are reluctant to regulate a substance about whose 
hazards the CPSC's scientific staff are convinced — perhaps because consumer 
exposures are higher '^^ — adopting a "me too" approach would disable it from 
regulating. In addition, such an approach would probably damage agency morale. 

c. Occupational Safety and Health Administration. Until the publication 
of its generic carcinogen policy in 1980, OSHA had no formal system for 
identifying candidates for occupational health standards. Petitions from labor 
unions had preceded the adoption of most^'^ of the health standards that OSHA 
adopted during the 1970's, although each petition had relied upon a criteria 
document previously prepared by the National Institute for Occupational Safety 
and Health (NIOSH). It was NIOSH, through preparation of criteria documents, 
that dictated OSHA's early agenda. '^^ By 1976, however, the number of un- 
answered criteria documents so far exceeded the number of chemicals that OSHA 
could hope to regulate that NIOSH had ceased to influence the regulators' prior- 
ities. ^^^ 

OSHA's decision to set standards for such substances as cotton dust, coke 
oven emissions, benzene, arsenic were not the product of systematic review of 
occupational health hazards generally or even of the more limited number of 
hazardous substances identified by NIOSH. Each standard represents a response 
to unique pressures. ' '^ It would be hard to criticize OSHA's decisions to regulate 
most of these substances. There can be little doubt, for example, that cotton 
dust and coke oven emissions, to pick two of its most controversial targets, pose 
serious risks to exposed workers. ' ^^ But this does not answer the question whether 
another more systematic approach might have enabled the agency to regulate 
more substances during the same period or with the same expenditure of agency 
resources — and yielded greater net benefits to worker health. 

OSHA's vulnerability to charges that it was not regulating the "worst first" 
and that it was concentrating too much effort on too few substances partially 

109. The CPSC has voted to ban urea- formaldehyde foam insulation. See [1982] Chemical 
Regulation Reporter (BNA). In contrast, EPA has determined that the environmental and health 
risks posed by formaldehyde do not now justify regulatory action under TSCA. See Risk Assessment 
in the Federal Government 122-25. 

110. T. Bartman, Deciding What To Regulate: Priority-Setting at OSHA (unpublished paper 

111. Id. 

112. Id. See also McGarrity, The Occupational Safety and Health Administration' s Generic 
Carcinogen Policy: Rulemaking Under Scientific and Legal Uncertainty (unpublished paper prepared 
for National Center for Administrative Justice, April 1981). 

113. Bartman, supra. 

1 14. See OSHA Draft at 161-87. 


explain the decision to promulgate the agency's generic carcinogen policy. That 
policy was not only intended to define the criteria OSHA would follow in 
evaluating and setting exposure limits for individual chemicals, but to establish 
a system for developing candidates for new standards. The final January 1980 
regulations accordingly set forth a comprehensive system for setting priorities. ' '^ 

OSHA officials have reopened the rulemaking proceeding and asked for 
comments on several portions of the carcinogen policy."^ Whether and how the 
agency should set priorities are issues opened to reexamination, and the public 
ranking scheme described below has been stayed pending review of the com- 
ments."^ Despite debates about its future, OSHA's announced system for se- 
lecting chemicals for regulation is the most elaborate devised by any agency. 
To facilitate understanding of the controversy that this initiative has engendered, 
it is useful to outline its main features. 

OSHA's adoption of a formal process for screening substances and setting 
priorities was in part a response to public comments on its original proposed 
carcinogen policy."^ Several individual witnesses, organizations, and sister agencies 
urged OSHA to adopt such a system and to make it public. OSHA agreed with 
the argument that, to the extent practicable, it should regulate the "worst first." 
However, the agency also insisted that priority-setting should not be so elaborate 
that it depleted resources for setting standards."^ 

Under OSHA's regulations the first step would be publication of a so-called 
Candidate List. This list would consist of between 200 and 450 chemicals iden- 
tified, through review of NIOSH's own list and from other sources, as worthy 
of further study as potential human occupational carcinogens.'^^ While OSHA's 
review of scientific literature would be cursory at this stage, chemicals would 
not automatically be included based on NIOSH's recommendations. A chemical's 
appearance on EPA's Toxic Substances Control Act inventory would be accepted 
as indicating probable occupational exposure. '^' Evaluations would be performed 
internally by OSHA staff or by contractors with the agency. '^^ 

After more thorough review of chemicals on the candidate list, OSHA would 
make public two priority lists, each containing approximately ten chemicals, 
characterized as Category I and Category 11 potential occupational carcinogens. '^^ 

115. 45 Fed. Reg. 5001 (1980). 

1 16. 47 Fed. Reg. 187 (1982); see [1981] Chemical Regulation Reporter 560 (BNA). 

117. 47 Fed. Reg. 187 (1982). 

118. 45 Fed. Reg. 5002, 5005-07 (1980). 

119. /^. at 5011. 

120. /J. at 5207-11. 

121 . OSHA briefly addressed concerns, expressed by several commentators, that mere listing 
would stigmatize a chemical in the commercial world before any considered review. The agency 
dismissed this concern, emphasizing that the chemicals would be subjected to some evaluation before 
listing and by claiming that the term "candidate" should convey the list's highly preliminary 
character. See id. at 5209. 

122. 45 Fed. Reg. 53672 (1980). 

123. 45 Fed. Reg. 5002, 5206 (1980). 


The two categories represent different levels of conviction, based on the sup- 
porting scientific evidence, that a chemical is a carcinogen. These lists, to be 
updated every six months, would comprise the universe of chemicals that OSHA 
would subject to thorough internal review preparatory to initiating rulemaking. ^^"^ 
Before including a chemical on either Priority List, OSHA would consider seven 
criteria: (1) the estimated number of exposed workers; (2) the estimated levels 
of worker exposure; (3) the levels of exposure reported to cause an increased 
incidence of neoplasms in humans and/or animals; (4) the extent to which reg- 
ulation can reduce the risk of cancer; (5) the extent to which the substance has 
a molecular structure similar to that of another substance which meets OSHA's 
definition of a potential occupational carcinogen; (6) whether there is evidence 
that the social and economic costs of regulation would be small; and (7) the 
relationship of the substance to regulatory activities of other agencies. '^^ 

OSHA stressed that inclusion of a chemical on either list would not constitute 
a final agency determination that it was an occupational carcinogen, but merely 
a judgment that it was a likely candidate for regulation. ^^^ While new candidates 
for health standards would ordinarily be selected from one of the two Priority 
Lists, OSHA disclaimed an intention to following the order of the lists or to 
confine itself initially to Category I chemicals. '^^ It pointed out that the lists 
reflected considerations, such as ease of control and availability of substitutes, 
which might justify earlier attention for lower-ranking chemicals. '^^ 

It should be noted that OSHA's 1980 priority setting system does not 
contemplate rulemaking to select individual chemicals for regulation. While the 
agency invited the submission of data concerning chemicals on the Candidate 
List, it did not obligate itself to consider or reply to any material submitted in 
response to either the Candidate List or the Priority Lists. '^^ Essentially what 
OSHA's regulations describe is a system for periodically releasing the results 
of an internal review process. This has fostered complaints from industry or- 
ganizations that OSHA's process will "indict" chemicals in the public mind 
before manufacturers or users have an opportunity to be heard. '^^ A commitment 
to more formal public procedures, however, with the attendant obligations to 
listen and respond, would substantially increase the costs of priority setting. A 
wholly private process, which did not result in disclosure of the chemicals the 
agency was considering for regulation, would substantially curtail opportunities 
for public involvement prior to the standard-setting process itself. 

124. /^. at 5207-11. 

125. /^. at 5210-11. 

126. Obviously attempting to forestall legal challenges to the listing of a chemical, perhaps 
based on a failure to provide some sort of advance hearing, the agency asserted: "The Priority Lists 
are not intended to create any legal rights." Cf. Dow Chemical, USA v. Consumer Product Safety 
Commission, 459 F. Supp. 378 (W.Df La. 1978). 

127. 45 Fed. Reg. 5210-11 (1980). 

128. Id. at 5210. 

129. Id. at 5208. 

130. See, e.g., id. at 5209. See also note 79 supra. 


d. Environmental Protection Agency. ^^^ EPA's Office of Pesticide Pro- 
grams has been engaged in three important activities for which it sets priorities. 
The agency is responsible for evaluating new pesticides submitted for registration; 
it has devoted considerable effort to the development of so-called generic stan- 
dards for regulating marked pesticides; and it has evaluated a comparatively 
small number of registered pesticides under the Rebuttable Presumption Against 
Registration (RPAR) process. The latter program, on which EPA's efforts to 
assess the carcinogenicity of pesticides have been concentrated, is a separate 
chapter. Some discussion of EPA's techniques for setting priorities for all three 
activities will illustrate how context and purpose influence this generic function. 

The sequence of EPA's evaluation of new pesticides is dictated externally 
by registration applications from manufacturers and distributors. The universe 
of new compounds presented to the agency is thus a function solely of private 
judgments. The agency generally processes applications for registration in the 
order received, occasionally considering other criteria, such as potential contri- 
butions to agricultural production or reductions in risk, in scheduling applications 
for review. *^^ 

EPA's registration standards program is an effort to develop generic criteria 
describing the characteristics of products containing similar active pesticide in- 
gredients.'-^^ A registration standard is a monograph, issued through rulemaking, 
which establishes the conditions of approval for products containing a specific 
pesticide. Promulgation of a standard requires EPA to confirm that toxicological 
and environmental data concerning the pesticide's properties support the judg- 
ment that it will not cause ''unreasonable adverse effects."'-^'* Once a standard 
has been established, manufacturers of products containing the pesticide can 
secure registration by demonstrating that their products conform to the standard. 
EPA has been developing these generic standards for several years, but its 
progress has been slow for several reasons. Notable among these are the vast 
number of registered products that EPA must examine and the relatively meager 
scientific data on which to base contemporary judgments about their safety. '^^ 

EPA has developed informal criteria for selecting pesticide ingredients as 
candidates for generic standards. '^^ These criteria emphasize the scope of pes- 
ticide use and potential human and environmental exposure, but priorities are 
influenced also by the adequacy of available data on health and environmental 
effects. The agency has no public system for selecting candidates for registration 

131 . See generally, NRG Regulating Pesticides (1980). 

132. However, applications which come within already existing registration standards can be 
acted on promptly; those that require additional scientific evaluation or testing inevitably take longer 
before final approval or rejection. Telephone interview with William Dickinson, Deputy Director, 
Special Pesticide Review Division, EPA (November 25, 1981). 

133. See 44 Fed. Reg. 7631 1 (1979). See also 45 Fed. Reg. 72948 (1980). 

134. FIFRA § 3(c)(5), (g), 7 U.S.C. 136a(c)(5)(g) (1976). 

135. See General Accounting Office, Delays and Unresolved Issues Plague New Pesticide 
Protection Programs 4-27 (1980). Interview with William Dickinson, supra note 134. 

136. Interview with Janet Auerbach, Director, Regulatory Support Branch, Special Pesticide 
Review Division, EPA (July 13, 1981). 


Standards; these criteria govern internal decisionmaking by the Office of Pesticide 
Programs division responsible for developing proposed standards. '^^ 

The third and, in the present context, most important OPP activity is the 
RPAR program. This is the mechanism by which EPA formally evaluates the 
risks and benefits of pesticides suspected of posing serious health hazards, in- 
cluding cancer. '''^ When it launched the RPAR review process in 1975,'^^ EPA's 
Office of Pesticide Programs had compiled a list of some 40 pesticide ingredients 
that it regarded as potential candidates for cancellation or curtailed use, primarily 
because of their potential for causing chronic disease. This list was drawn up 
by the agency staff without public participation. A variety of criteria influenced 
their decisions to include a pesticide. The core of the list, perhaps 25 pesticides, 
came from an earlier listing of "suspect" pesticides based on a 1969 study 
submitted to the Department of HEW — the so-called Mrak Commission Re- 
port. '"^^ Other pesticides were added to the initial list of RPAR candidates fol- 
lowing hasty review of safety data in EPA's files and examination of approved 
food tolerances, which were assumed to be a proxy measure of the extent of 
human exposure. ^"^^ 

EPA's original universe of RPAR candidates was, thus, drawn from an 
earlier list of "problem pesticides," supplemented by cursory review of infor- 
mation in the agency's files. According to one EPA official, the pesticides on 
the list were ranked by the staff according to their toxic potency, their persistence 
in the environment, total production volume, and a rough assessment of their 
agricultural importance. '"^^ This informal ranking has not, however, dictated the 
sequence in which EPA has evaluated the pesticides; the agency has often moved 
ahead with substances it considers easy to resolve — either by continued approval 
or by cancellation. The timing of RPAR actions has also been heavily influenced 
by external events, including criticism by two Congressional committees and 

137. Id. 

138. Office of Pesticide Programs, EPA, June 1981 Status Report on Rebuttable Presumption 
Against Registration (RPAR), Registration Standards, and the Data Call-in Program (1981). 

139. 40 Fed. Reg. 26802 (1975). 

140. U.S. Department of Health, Education, and Welfare, Report of the Secretary's Com- 
mission on Pesticides and their Relationship to Environmental Health, Parts I and II (1969). Several 
of the problem pesticides that the report identified were eventually suspended or cancelled by EPA 
before commencement of the RPAR review process. See Environmental Defense Fund, Inc. v. 
Ruckelshaus, 439 F.2d 584 (D.C. Cir. 1971) (DDT); EDF v. EPA, 510 F.2d 1292 (D.C. Cir. 1975) 
(aldrin/dieldrin); EDF v. EPA, 548 F.2d 998 (D.C. Civ. 1976) (heptachlor/chlordane). 

141 . Interview with Steven Jellinek, Former Assistant Administrator for Pesticides and Toxic 
Substances, EPA (June 23, 1981). 

142. Telephone interview with William A. Wells, Special Assistant to the Assistant Admin- 
istrator for Pesticides and Toxic Substances, EPA (July 10, 1981). 


reports of epidemiological studies purporting to demonstrate adverse health ef- 
fects among exposed humans. '"^^ 

A 1980 report by the National Academy of Sciences, which studied the 
RPAR process, criticized EPA for failure to establish a formal, hazard-based 
system for setting priorities. '"^"^ Whether or not prompted by this report, by 1981 
OPP had formed an internal planning group, headed by the Deputy Assistant 
Administrator for Pesticide Programs, division heads, and representatives of 
EPA's Office of General Counsel to select candidates for regulatory action.''*^ 
Such action could be inclusion in the registration standards process or scheduling 
for RPAR review. At this time, however, EPA does not contemplate publishing 
formal criteria for candidate selection or establishing a system that permits public 
participation in the selection process. ^"^^ 

D. Institutional Mechanisms for Assessing Risk: Advisory Panels 

This section and the next one examines a variety of procedures by which 
regulatory agencies can evaluate the capacity of chemicals to cause cancer. The 
discussion betrays a conspicuous omission: It does not systematically consider, 
except by contrasting example, the method most used by regulatory agencies, 
namely, having staff scientists review the available data and report their con- 
clusions to the agency head. This approach is always available, although its 
adequacy of course depends on the scientific capabilities of agency staff. It is 
also the approach likely to be followed in routine cases by all of the agencies, 
particularly EPA and FDA, which receive a continuous stream of approval 
applications for new products, new uses for old products, residue tolerances, 
and similar matters requiring some evaluation of potential human risk. 

The focus of these sections is on institutional methods for assessing health 
risks in major or controversial cases. This section deals with one important 
approach with which the agencies have had considerable experience, the use of 
expert advisory panels. Although treated generically, it should be evident from 
the discussion that advisory panels assume different forms and function in dif- 
ferent ways. The following section examines recent proposals for new institu- 
tional approaches to risk assessment. 

1 . Introduction 

Advisory panels warrant special attention for several reasons. First, the 
current agency practice of consultation with expert committees differs in im- 

143. See, e.g.. Staff of Senate Subcomm. on Administrative Practice and Procedure of the 
Comm. on the Judiciary, 94th Cong., 2d Sess., Report on the Environmental Protection Agency 
and the Regulation of Pesticides (Comm. Print 1976); Effects of 2,4,5 -T and Related Herbicides on 
Man and the Environment: Hearings Before the Senate Subcomm. on Energy, Natural Resources, 
and the Environment of the Comm. on Commerce, 91st Cong., 2d Sess. (1970); Staff of House 
Subcomm. on Oversight and Investigations of the Comm. on Interstate and Foreign Commerce, 
94th Cong., 2d Sess., Report on Federal Regulation and Regulatory Reform 111-51 (1976). 

144. Regulating Pesticides at 46-96. 

145. Auerbach interview, supra note 138. 

146. Id. 


portant ways from some proposals to establish a new body for resolving scientific 
questions, which emphasize separation from agency regulators, centralization, 
and authoritative resolution. '^^ As described here, advisory panels often function 
as parts of an agency's decisionmaking apparatus, frequently serve a single 
agency, and provide only advice, not binding decisions. Panel judgments may 
not be accepted by the agency and, even when they are, theoretically are open 
to challenge by private parties in whatever formal process the agency must 
provide. A second reason for treating advisory panels separately is that the four 
agencies have had a good deal of experience with them. It is therefore possible 
to describe their operation and to evaluate their performance empirically rather 
than hypothetically. Furthermore, examples of the successful use of advisory 
panels provide guidance for the design of any new institutions to assess carcin- 
ogenic risks. 

2. Rationale for Agency Use of Advisory Panels. 

For purposes of this discussion an "advisory panel" is any group of ex- 
perts — whether or not government employees — assembled by an agency to aid 
it in evaluating the health risks posed by a product or activity. This definition 
is broader than that of the Federal Advisory Committee Act (FACA),^"^^ which 
is addressed to formally chartered committees consisting predominantly of non- 
government members.''*^ 

While the structure and procedures of agency advisory panels have varied, 
their use reflects a common rationale: Agencies have found it desirable to obtain 
the focused views of outside experts on the health risks posed by chemicals that 
are candidates for regulation. Several advantages have been perceived in such 
consultation. First, it is sometimes assumed that agency scientists lack the training 
to provide reliable judgments; in this view, advisory panels supply expertise that 
the agencies lack.^^^ Second, an agency may seek support from the scientific 
community to bulwark conclusions reached by its own staff; advisory panels can 
thus enhance the credibility of agency judgments. ^^' Third, consultation with an 

147. See, e.g.. National Science Council Act of 1980, H.R. 638, 97th Cong., 1st Sess., 127 
Cong. Rec. 2, 67 (1981). This bill, popularly known as the Wampler Bill in recognition of its 
sponsor, was originally introduced as H.R. 6521, 96th Cong., 2d Sess., 126 Cong. Rec. 22, 901- 
02 (1980). Another proposal for the creation of a science panel has been made by the American 
Industrial Health Council. "AlHC Proposal for a Science Panel," American Industrial Health 
Council, March 26, 1980. See generally. Risk Assessment in the Federal Government. 

148. Pub. L. No. 92-463, 86 Stat. 770 (1972), codified at 5 U.S.C. Appendix I (1976). 

149. Agencies have also employed other mechanisms for obtaining guidance and information 
from outside experts, including contracts with private firms and research organizations and appoint- 
ment of individual consultants. 

150. This is a conclusion reached by the Office of Technology Assessment in chapter 9 of 
an as yet unpublished study of the relationship of technological innovation and regulation. See also 
Merrill and Hutt at 907-08. 

151. EPA's Scientific Advisory Panel appears to have strengthened the credibility of the 
agency's regulation of controversial pesticides by subjecting EPA's own hazard evaluations to 
disinterested expert review by a panel of nongovernmental scientists. Interview with Steven Jellinek, 
former Assistant Administrator for Pesticides and Toxic Substances, EPA (June 23, 1981); interview 
with Philip Gray. Executive Secretary, Scientific Advisory Panel, EPA (July 14, 1981). 


advisory panel may afford an agency time to explore regulatory options or permit 
industry to develop substitutes, thereby cushioning the impact of regulation. '^^ 
Fourth, consultation with outside experts can dilute the effect of policy biases 
that, it is believed, often influence agency risk assessments. On this premise, 
advisory panels contribute objectivity. '^-^ 

One of the most important justifications for the use of advisory panels in 
assessing chemical hazards is that it institutionalizes ''peer review" of laboratory 
findings, which epitomizes scientific inquiry. As Dr. William Lowrance has 

. . . [I]t is only through a rather slow process of cumulative criticism and 
endorsement by the diaspora of scientists that science grows. Science is, 
in effect, simply a matter of "voting," with re-balloting conducted from 
time to time. The scientifically True is no more than what scientists endorse 
to be true at the moment. Empirical knowledge is developed systematically 
within the scientific community, subject to criteria of repeatability, con- 
trolled observation, statistical significance, openness, and the other guides 
of Western science. Procedure by itself guarantees nothing, though. Good 
science is science that "works": science that can predict with consistency 
and generality and accuracy what will happen in the physical and social 
world. Weighing of the compellingness of candidate "facts" remains sub- 
jective; perfect objectivity is a myth. 

Consequently, we must depend heavily on committees and other struc- 
tured groups of scientists to screen the quality of science used in public 
decisionmaking. . . .'^"^ 

3. Approaches to the Use of Advisory Panels. 

A survey of agency experience with advisory panels reveals a variety of 
forms, procedures, and authorities. This diversity may contribute to suspicions 
that agencies devise a new approach for each new problem.'-'^'' The impression 
that regulatory agencies "make up the rules as they proceed" may justify de- 

1 52. FDA bought considerable time by referring the issue of saccharin to the National Academy 
of Sciences in 1968 and again in 1972. See R. Merrill and M. Taylor, saccharin (unpublished 
manuscript, March 1983). 

153. See, e.g., S. Rep. No. 94-452, 94th Cong., 1st Sess. reprinted in [1975] U.S. Code 
Cong. & Ad. News 1358. for suggestions that EPA's Scientific Advisory Panel would enhance 
objective scientific analysis within the agency's decisionmaking apparatus. Much the same rationale 
underlies FDA's use of drug review committees. See Merrill and Hutt, supra at 907-09. 

154. See Risk Assessment in the Federal Government at 127. W. Lowrance, Better Science 
for Regulatory Decisionmaking, pp. 12-13 (Paper delivered to a Brookings Institution Conference, 
July 23, 1981). 

155. A central recommendation by the General Accounting Office, following a study of FDA's 
investigations of possible health hazards caused by nitrites, was that FDA should develop more 
standardized procedures for risk assessment. GAO, Does Nitrite Cause Cancer? Concerns About 
Validity of FDA-Sponsored Study Delay Answer (Jan. 31, 1980). A similar criticism of past FDA 
performance appears in a Citizen Petition submitted to FDA by the Grocery Manufacturers of America 
in August, 1981. 


velopment of some standard procedures, but it should not obscure the importance 
of allowing agencies to experiment with new approaches to achieve the basic 
objective — the introduction of independent scientific judgment into agency de- 

a. Standing v. ad hoc committees. An agency may establish, or be required 
to establish, a single panel to review all its risk assessments. For example, EPA's 
Office of Pesticide Programs (OPP), is required by statute to consult with a 
Science Advisory Panel (SAP) on major questions concerning the hazards of 
marketed pesticides J -^^ At the other extreme is an advisory panel created for the 
purpose of reviewing a single study or a single chemical. ^^^ The former approach 
provides continuity and facilitates consistency; the latter approach permits the 
selection of panelists who possess precisely the expertise needed to evaluate the 
chemical in question. 

b. Source of members . Standing panels which advise regulatory agencies 
typically consist of scientists drawn from outside the government, most from 
the academic community. Panels assembled to consider a specific chemical, by 
contrast, have often been composed of government scientists, many of them 
employed at one or more of its major research facilities, such as NCI, the National 
Institute of Environmental Health Sciences (NIEHS), or the National Center for 
Toxicological Research (NCTR). Use of non-government scientists generally 
enlarges the universe of experts qualified for service and perhaps contributes to 
the appearance of objectivity. But it also increases cost and usually prolongs the 
consultative process, for the panel then becomes subject to the requirements of 
the F AC A. '^^ 

c. Procedural context. Advisory panels have been consulted at different 
stages of the regulatory process and in both rulemaking and adjudicatory pro- 
ceedings. Probably the most common practice is for an agency to consult a panel 
before initiating formal regulatory action, e.g. , by publishing a notice of proposed 
rulemaking. In such a case, the committee's findings and its report, if any, 
become part of the announced basis for the agency's proposal and are subject 
to challenge in the public administrative process. But an agency may also use 
an advisory committee after it has initiated regulatory proceedings, perhaps in 
response to criticism of its initial proposal. This latter approach requires careful 
scheduling to assure that the public has an opportunity to comment on the panel's 
findings. Such a panel may even function as the agency's tribunal for hearing 
and evaluating public comments. 

d. Scope of inquiry. Agencies have used expert panels to evaluate the risk 
of individual chemicals based on all available data, but they have also formed 
ad hoc panels to evaluate single studies. This latter approach has usually been 

156. See FIFRA §§ 6(b), 25(d), 7 U.S.C. §§ 136d(b), 136w(d) (1976). 

157. At the CPSC's request, the National Toxicology Program established such a panel to 
evaluate data on formaldehyde, while FDA created a panel devoted to the assessment of Red No. 
40. See Risk Assessment in the Federal Government 122-25. 

158. Id. 


employed only where a chemical has long been a subject of controversy and its 
effects — apart from the findings of the new study — are well-described. The 
questions submitted to an advisory panel can range from the narrowly scientific — 
"Does this latest study indicate that saccharin causes cancer in rats?"'^^ — to 
the broad and policy-dominated — "Considering the experimental data, should 
this substance continue to be marketed?" Several FDA committees have been 
asked to answer questions of the latter type for groups of related products.'^ 

The questions submitted to a panel must be framed to elicit answers that 
permit an agency to reach a decision, i.e., whether or not to restrict exposure, 
require more testing, or ignore a chemical. Yet in attempting to structure a 
panel's response an agency runs two risks. First, it may invite charges that it 
tampered with the panel's independence. Second, it may introduce into the 
panel's deliberations legal or policy issues that lie beyond its competence. Despite 
precautions, however, it is not always possible to confine a panel to the issues 
posed by the agency.'^' Panel members may regard those issues as incorrect or 
they may not be content to confine their advice to the "purely" scientific ques- 
tions that occasion their appointment. 

e. Intra-agency v. interagency panels. Most panels have been appointed 
by, and served, a single agency. But there are no insurmountable legal obstacles 
to creation of interagency expert panels to evaluate chemicals of common con- 
cern. In one recent instance, the National Toxicology Program (NTP), assembled 
a panel of government scientists to evaluate a chronic study of formaldehyde.'^^ 
It was anticipated that the conclusions of this panel would provide a common 
starting place for all four agenices with jurisdiction over uses of the substance. '^^ 

4. Determinants of Agency Approach 

Consultation with advisory panels has not been a routine part of agency 
evaluation of chemical risks; agencies have generally relied exclusively on their 
own scientific staffs. For example, OSHA has adopted five health standards 
since 1977 without once invoking its authority to establish an ad hoc advisory 
committee. '^ With increasing frequency, however, the four agencies have sought 

159. This was the assignment to the NTP panel on formaldehyde. See note 155, supra, and 
accompanying text. 

160. See, e.g., OTC Drug Review, 21 C.F.R. § 330.10 (1981), and the Biologies Review, 
21 C.F.R. § 601.25 (1981); Merrill and Hutt at 439-52, 520-28. 

161. EPA's Scientific Advisory Panel, for instance, has occasionally exceeded its statutory 
authority to review pesticide health and environmental risks by also considering corresponding 
benefits of continued pesticide use and specific regulatory options. Interview with Philip Gray, 
Executive Secretary, Scientific Advisory Panel, EPA (July 14, 1981). 

162. 46 Fed. Reg. 1 1 190 (1981). See Merrill: Consumer Products at 1355. 

163. Interview with Dr. Peter Preuss, CPSC Associate Executive Director for Health Sciences 
(December 15, 1981). 

164. Interview Charles Gordon, Office of the Solicitor, Department of Labor, June 23, 1981 . 
See generally. Risk Assessment in the Federal Government at 91-93. 


the advice of outside experts and explored ways of institutionalizing scientific 
"peer review." Several considerations have inspired their interest in independent 
scientific advice. 

Several statutes mandate agency consultation with advisory panels. Since 
1975, FIFRA has required OPP to submit proposals to cancel or restrict pesticides 
to a Science Advisory Panel (SAP), a committee whose members are appointed 
by the EPA Administrator from nominees submitted by the National Academy 
of Sciences and National Science Foundations.'^^ Recent amendments to the 
Consumer Product Safety Act require the CPSC to establish a Chronic Hazard 
Advisory Panel to review the scientific data on any chemical for which it con- 
templates a standard or ban.'^^ The OSH Act does not demand, but it expressly 
permits the Assistant Secretary for Occupational Safety and Health to establish 
committees to provide advice on the risks posed by specific workplace hazards. '^^ 
OSHA consulted with such committees before promulgating standards for several 
chemicals. '^^ 

Though the issue has never been litigated, it seems clear that, even without 
statutory authorization, an agency may seek the advice of independent experts 
by retaining them as consultants, chartering an advisory committee, or consulting 
with those employed elsewhere in government. Accordingly, without so much 
as a whisper of Congressional approval, FDA has established the largest entou- 
rage of expert advisory committees serving any regulatory agency. These com- 
mittees have played a central role in FDA's efforts to evaluate the safety and 
effectiveness of health care products, including prescription and OTC drugs, and 

165. FIFRA § 25(d), 7 U.S.C. 136w(d) (Supp. IV 1980) provides in pertinent part: 

The Administrator shall submit to an advisory panel for comment as to the impact on 
health and the environment of the action proposed in notices of intent issued under section 
1 36d(b) of this title and of the proposed and final form of regulations issued under subsection 
(a) of this section within the same tim.e periods as provided for the comments of the 
Secretary of Agriculture under such section I36d(b) and subsection (a) of this section. 
. . . The Administrator shall also solicit from the advisory panel comments, evaluations, 
and recommendations for operating guidelines to improve the effectiveness and quality of 
scientific analyses made by personnel of the Environmental Protection Agency that lead 
to decisions by the Administrator in carrying out the provisions of this subchapter. The 
comments, evaluations, and recommendations of the advisory panel and the response of 
the Administrator shall be published in the Federal Register in the same manner as provided 
for publication of the comments of the Secretary of Agriculture under such sections. . . . 
Whenever the Administrator exercises authority under section I36d(c) of this title to 
immediately suspend the registration of any pesticide to prevent an imminent hazard, the 
Administrator shall promptly submit to the advisory panel for comment, as to the impact 
on health and the environment, the action taken to suspend the registration of such pesticide. 

166. Consumer Product Safety Amendments of 1981, Pub. L. No. 97-35, § 1206, 95 Stat. 
357 (1981) (amending 15 U.S.C. § 2077 (1976)). The desirability of consultation with expert 
scientific organizations was a topic of interest in the debate that led to the Medical Device Amend- 
ments of 1976. See Study Group on Medical Devices, Medical Devices: A Legislative Plan (DHEW, 
1970), reprinted in Merrill and Hutt at 577. 

167. OSH Act, § 7, 29 U.S.C. § 656 (1976). 

168. OSHA Draft at 22, 137. 


biological products. ^^^ Most of these committees were created to perform func- 
tions, e.g., the review of pre- 1962 drugs, that were beyond the capacity of 
FDA's regular staff. '^^ 

5. The Federal Advisory Committee Act. 

Passed in 1970^^' and later amended by the Government in the Sunshine 
Act,'^^ the FACA was intended to regularize the procedures of the several 
hundred advisory committees established by the federal government during the 
1950's and 1960's. A companion objective was to curtail agency reliance on 
outside advisers who were involved with or employed by regulated enterprises. 
The origins of the FACA need not be explored in depth here, but it is worthwhile 
to highlight the Act's basic requirements for creating and operating advisory 
committees. These requirements not only govern the behavior of bodies covered 
by the Act but influence the manner in which agencies elicit scientific advice. 

The FACA does not either require or prohibit the creation of advisory 
committees; it thus does not determine when an agency may consult with outside 
experts. The Act does, however, impose significant requirements for the estab- 
lishment and operation of bodies that meet its definition of "advisory commit- 
tee." That broad definition '^^ encompasses many of the advisory panels employed 

169. 5^^ Merrill and Hutt 907-919. 

170. Id. See also Note, Drug Efficacy and the 1962 Drug Amendments, 60 Geo. L.J. 185 

171. Pub. L. No. 92-463, 86 Stat. 770 (1972), codified at 5 U.S.C. Appendix I (1976). 

172. Pub. L. No. 94-409, § 5(c), 90 Stat. 1247 (1976), codified at 5 U.S.C. Appendix I § 
10(d) (1976). The brief amendment to the FACA included in the Sunshine Act was intended to 
ensure that advisory committee meetings remain predominantly open to the public. Original section 
10(d) of FACA provided that "Subsections (a)( 1) and (a)(3) [requiring open meetings and permitting 
public participation] shall not apply to any advisory committee meeting which the President, or the 
head of the agency to which the advisory committee reports, determines is concerned with matters 
listed in section 552(b) of title 5, United States Code." Amended section 10(d) of FACA reads as 
follows: "Subsections (a)(1) and (a)(2) of this section shall not apply to any portion of an advisory 
committee meeting where the President, or the head of the agency to which the advisory committee 
reports, determines that such portion of such meeting may be closed to the public in accordance 
with subsection (c) of section 552b of title 5, United States Code." 

173. Section 3(2) reads as follows: 

The term "advisory committee" means any committee, board, commission, council, 
conference, panel, task force, or other similar group, or any subcommittee or other subgroup 
thereof (hereafter in this paragraph referred to as "committee"), which is — 

(A) established by statute or reorganization plan, or 

(B) established or utilized by the President, or 

(C) established or utilized by one or more agencies. 

in the interest of obtaining advice or recommendations for the President or one or more 
agencies or officers of the Federal Government except that such term excludes (i) the 
Advisory Commission on Intergovernmental Relations, (ii) the Commission on Government 
Procurement, and (iii) any committee which is composed wholly of fulltime officers or 
employees of the Federal Government. 


by FDA, EPA, OSHA, including most committees whose existence is mandated 
by other laws, such as FIFRA's Science Advisory Panel. '^"^ 

The Act's application to informal arrangements by which government of- 
ficials obtain advice has been at issue in several cases. ^^^ The courts have 
construed the Act broadly to embrace most agency efforts to obtain advice from 
nongovernment experts. ^^^ A decision of the Second Circuit, for example, holds 
that FDA officials' consultation with several academic experts on nutrition who 
were attending a professional meeting made the group an "advisory committee" 
under the FACA, and thus subject to its requirements for chartering, public 
notice of meetings, and open deliberations.'^^ It is these requirements, though 
not individually onerous, that together have sometimes discouraged agencies 
from soliciting scientific advice or induced them to find other means of exposing 
staff assessments to outside peer review. '^^ 

The FACA requires that any advisory committee must be formally chartered 
by the agency head, with the concurrence of the General Services Administra- 
tion.'^^ A continuing committee must be rechartered every two years. '^^ While 
simple on its face, this procedure has often proved cumbersome. Some agencies 

174. See note 162, supra, the CPSC's advisory panels drawn from other governmental agencies 
are explicitly excluded from FACA's requirements under § 3(2) (C) (iii) of FACA. And under the 
1981 amendments to the CPS Act, the CPSC's Chronic Hazard Advisory Panel is explicitly excepted 
from coverage of the Federal Advisory Committee Act. 

175. See, e.g.. National Nutritional Foods Ass'n v. Califano, 603 F.2d 327 (2d Cir. 1979); 
Consumers Union of United States, Inc. v. Department of HEW, 409 F. Supp. 473 (D.D.C. 1976), 
affd without opinion, 551 F.2d 466 (D.C. Cir. 1977). 

176. Id. 

177. National Nutritional Foods Ass'n v. Califano, 603 F.2d 327 (2d Cir. 1979) (Friendly, 
J.). About this decision Judge Friendly has written: 

No one seems even to have considered that in seeking to cure one trouble by imposing 
procedural requirements, the statute would create others. Just what is an advisory com- 
mittee? In a recent opinion, where the FDA had consulted with a number of physicians 
who were attending a meeting of the nearby National Institute of Health, I regretfully 
found it impossible to accept the Government's position that simply by keeping things 
informal an agency can escape from a statute whose very purpose is to require formality. 

Against this a judge would scarcely wish to say that if on the night of Three Mile Island 
the NRC wanted to telephone a few eminent atomic scientists, it could not legally have 
done so. Assuming that a talk with one scientist would not have been subject to FACA, 
what about a conference call with five? What about a couple of meetings in Harrisburg 
during the next few days? Beyond this the statute necessitates additional staff in each 
agency and in the Office of Management and Budget. Would we not have been better off 
with a bit more trust and less law? 

Friendly, Should We Be Turning Back the Law Flood?, Legal Times of Washington, Oct. 8, 1979, 
at 7. 

178. A notable illustration of such alternative peer review is the formation in 1980, at the 
CPSC's initiative, of a panel of scientific experts from other government regulatory and research 
agencies to review existing data on formaldehyde and assess the carcinogenic potential of formal- 
dehyde gas leaking from urea formaldehyde foam insulation. 

179. FACA § 9. 

180. FACA § 14. 


lack independent authority to charter committees because department heads have 
been unwilling to relinquish control over their number and composition.'^' Fur- 
thermore, central procedures for screening new committees have imposed lengthy 
delays, often inspired by political concerns about committee membership or any 
apparent growth in the number of new federal entities. '^^ For example, FDA 
has had difficulty securing approval for new advisory committees since 1976. 

Once formally chartered, an advisory committee must comply with two 
requirements designed to assure that advice is provided on the public record. 
The first of these is the requirement that all meetings be announced in the Federal 
Register at least 15 days in advance, accompanied by a description of the subjects 
to be discussed. '^^ Desirable in principle, this requirement nonetheless inhibits 
agencies from seeking outside scientific evaluations when prompt action is re- 

The second requirement, clarified by the Sunshine Act,'^"^ is that meetings 
must be open to the public except in narrowly defined circumstances which 
would rarely justify closure of a committee empaneled to evaluate a chemical's 
risks. This open meeting requirement has proved less disruptive than some 
agencies initially anticipated,'^^ but it has discouraged some scientists from 
serving on advisory committees. Most members of expert committees are aca- 
demics, who are accustomed to debating experimental findings unencumbered 
by cameras, microphones, reporters, and vocal industry and environmental 
spokespersons. Such intrusions do impede frank exchange or divert it to ex- 
changes outside formal meetings. Allowing committees to meet in private when- 
ever they desired, however, would have other costs. 

These FACA requirements have not seriously impeded the work of standing 
advisory committees, whose responsibility and continuity make compliance a 
small price for expert advice. '^^ But the Act has frequently caused agencies to 
find other ways of obtaining peer review when the issues before them involve 
a single chemical or single test. In many such instances, regulators have confined 
their consultations to other government scientists, who can be made accessible 
immediately and, if necessary, for extended periods. '^^ 

One should not discuss the legal restrictions on advisory panels without 
noting the impact of current federal conflict of interest laws. Members of formal 
advisory committees are so-called "special government employees," who are 

181. FLA, for example, must obtain approval from the Assistant Secretary for Health and 
the Secretary of Health and Human Services. 

182. When President Carter entered office, departments and agencies were directed to cut 
substantially the number of existing advisory committees. 

183. FACA § 10. 

184. See note 172 supra. FDA's earlier use of advisory committees probably violated this 
provision of the FACA as amended. 

185. 5^6- Merrill and Hutt at 918. 

186. A useful example is EPA's Scientific Advisory Panel, which is discussed at texts ac- 
companying notes 264-302 infra. 

187. The CPSC followed this procedure with respect to formaldehyde, as did FDA in its 
evaluation of Red No. 40. 


subject to same post-employment restrictions as full-time government officials J ^^ 
These restrictions have discouraged some outside scientists from accepting ap- 
pointment to advisory committees. Perhaps more damaging to the government's 
ability to secure advice are the rules that forbid a committee member to participate 
in any matter in which he may have an interest J^^ This limitation, in principle, 
seems appropriate, but the Department of Justice's interpretation of the phrase, 
"particular matter," has extended the statute's prohibition to contexts where the 
potential for personal gain is remote '^"^ In addition, it is difficult to determine 
what kinds of financial interests are disqualifying, particulalry for academic 
scientists whose students, colleagues, and employers are often the recipients of 
research support from firms with an interest in products on which the panel's 
advice may be sought. ^^^ 

6. Agency Experience with Advisory Panels 

Each of the four agencies studied has on occasion sought the advice of 
expert panels about the health risks posed by specific chemicals. The following 
discussion does not attempt to describe each agency's experience comprehen- 
sively. Rather, it selects specific episodes to illustrate the advantages and lim- 
itations of this technique for aiding regulatory decisionmaking. 

a. Food and Drug Administration. FDA's Bureau of Foods has noi es- 
tablished a systematic approach to the use of outside experts in evaluating the 
safety of food constituents,'^^ but it has consulted advisory panels on most 
controversial decisions. In addition, other agency bureaus have for many years 
relied on expert committees for advice on the composition and labeling of health 
care products. In testimony before Congress in 1974, FDA Commissioner Schmidt 
gave several reasons for the agency's use of advisory committees: 

1. The agency gains access to the highest levels of scholarship in the 
scientific community and in Federal agencies other than the FDA. . . . 

2. Regulatory decisions are recognized by affected parties as having the 
backing of leaders in the medical, academic, and scientific communities, 
all of whom are visible and accountable. . . . 

4. In areas where the Agency lacks intramural competence to deal with a 
matter in a definitive fashion, participation of selected technical committees 
or panels permits a prompt and responsive effort. . . . 

188. See Merrill, Problems Involving Federal Conflict of Interest Restrictions on Members 
of FDA Advisory Committees and Agency Officials, University of Rochester Center for the Study 
of Drug Development Publication Series (1980). 

189. Id. 

190. Id. 

191. Id. 

192. See Citizen Petition to FDA by Grocery Manufacturers of America, Inc., August 20, 
1981, at 76-81. 


6. Exposure of consultants and committee members to the Agency's de- 
liberations and problems promotes a desirable dissemination of information 
which might otherwise be confined to official circles. . . . 

7. The incorporation of advisory committees into the review process aids 
decisionmaking by providing a formal setting for comprehensive review of 
data, scientific discussion, and resolution of problems.''^'' 

The importance of advisory panels to FDA is illustrated by their role in the 
agency's regulation of human drug products. In 1966 the National Acadamy of 
Sciences, under contract with FDA, assembled some thirty expert panels to 
review the effectiveness of pre-1962 prescription drugs. '*^'* In 1972, the agency 
launched a comprehensive review of the safety and effectiveness of all OTC 
drugs, in which advisory committees were given the central role.'^^^ One year 
later the agency established six additional expert panels to evaluate all human 
biologicals.'^^ The pace of these programs has been slow and their cost to FDA 
has been substantial, but both pale beside the costs and delays the agency would 
have encountered if it had attempted these review functions with full-time per- 
sonnel. Indeed, it is doubtful that FDA could have performed these tasks using 
only agency personnel. 

Each of these product reviews has assigned individual advisory panels to 
evaluate all of the products or ingredients in a single class. In addition to rendering 
"verdicts," the OTC drug and biological products panels have also been asked 
to produce fully-documented reports setting forth their findings and reasons. 
Successive FDA Commissioners have accepted the vast majority of their rec- 
ommendations and, though implementation of these recommendations has been 
slow, none has been challenged in formal proceedings before the agency or in 
court. One explanation for this acquiescence is that FDA has generally allowed 
time for manufacturers to substitute new ingredients or revise labeling. 

For several reasons, judgments of the panels assembled by NAS to review 
pre-1962 prescription drugs have been contested more frequently. The members 
of those panels, virtually all academics, remained anonymous throughout the 
review process. They met privately. And they produced merely verdicts, e.g., 
"ineffective," rather than documented reports. Moreover, before implementing 
judgments of the NAS panels FDA must, by statute, afford manufacturers of 
allegedly "ineffective" products an opportunity for a formal evidentiary hearing. 
Manufacturers of profitable products have thus had an incentive to insist on their 
full legal rights. '^^ 

193. Use of Advisory Committees by the Food and Drug Administration: Hearings Before a 
Subcomm. of the House Comm. on Government Operations, 93rd Cong., 2d Sess. 13-15 (1974). 

194. Merrill and Hutt at 908-09. 

195. Id. at 439 et seq. 

196. Id. at 520 et seq. 

197. For a detailed chronicle of FDA's efforts to implement the NAS-NRC findings, see J. 
Mashaw and R. Merrill, Introduction to the American Public Law System 459-542 (1975); 121- 
28 (1980 Supplement). 


FDA has been less systematic in seeking expert advice on the safety of 
substances in food.'*^*^ In 1974 the agency created a Toxicology Advisory Com- 
mittee''^'^ to which it would refer any scientific question arising under the Delaney 
clauses. ^^^ In practice, however, the agency rarely consulted this body^"' and 
when HEW Secretary Califano ordered a cut-back in departmental advisory 
committees in 1978, it was dissolved. 

During the past decade, FDA has followed several different approaches in 
evaluating the safety of food constituents. Often it has reviewed these issues 
internally. ^^^ On other occasions, it has requested review of an ingredient by 
the National Academy of Sciences (e.g., saccharin^^^), or by the Federation of 
American Societies for Experimental Biology. ^^"^ Once it sought advice from the 
National Cancer Institute (e.g., cyclamate^^^). The agency has also engaged 
outside experts as ad hoc consultants^^^ and ha5» convened informal working 
groups including both government and outside experts (e.g., FD&C Red No. 
4q207^ There is no evident rationale for the agency's choices among these di- 
vergent approaches. FDA's varied use of outside experts thus permits evaluation 
of several different mechanisms of peer review. The following discussion ex- 
amines three recent episodes in the agency's continuing efforts to regulate food 

Cyclamate. In 1969 FDA announced that cyclamate, then one of two avail- 
able artificial sweeteners^^*^ could no longer be considered "generally recognized 
as safe" because recent experiments indicated that it might cause cancer. This 
decision effectively banned the ingredient from food.^'^^ Subsequently, the man- 
ufacturer submitted a food additive petition seeking reapproval of cyclamate, 

198. The history of FDA's consultation with outside experts on the safety of substances in 
food is discussed in detail in the Citizen Petition by the Grocery Manufacturers of America, supra 
note 192, at 76-82. 

199. 39 Fed. Reg. 44473 (1974). 

200. 40 Fed. Reg. 40682, 40710 (1975), 21 C.F.R. 14.19. This regulation was inadvertently 
omitted from the republication of Part 14 in 44 Fed. reg. 22318, 22351, 22353 (1979) and thus also 
from the subsequent C.F.R. , but has not been revoked and remains in effect today. 

201. 43 Fed. Reg. 30271 (1978). 

202. See 37 Fed. Reg. 26618 (1972); 38 Fed. Reg. 20723 (1973); 43 Fed. Reg. 10551 (1978); 
39 Fed. Reg. 26748 (1974); 41 Fed. Reg. 19207 (1976). 

203. The National Academy of Sciences reviewed the safety of saccharin for FDA in 1955, 
1968, 1970, 1974, and 1978. National Academy of Sciences, Saccharin: Technical Assessment of 
Risks and Benefits 1-6 to 1-8 (1978). 

204. 38 Fed. Reg. 20053, 22054 (1973). 

205. 41 Fed. Reg. 43754 (1976). 

206. 38 Fed. Reg. 12642 (1973); 39 Fed. Reg. 17343 (1974). FDA's ad hoc effort to obtain 
advice from consultants on the safety of protein supplements was held to violate the Federal Advisory 
Committee Act. National Nutritional Foods Ass'n v. Califano, 603 F.2d 327 (2d Cir. 1979). 

207. 42 Fed. Reg. 8005 (1977); 42 Fed. Reg. 61630 (1977); 43 Fed. Reg. 18258 (1979); 44 
Fed. Reg. 30437 (1979). 

208. Saccharin was the other available artificial sweetener. See 42 Fed. Reg. 19996 (1977). 

209. Plotkin, The Cyclamate Ban: Science, Politics, and Law (April 1978) (unpublished M. A. 


arguing that the earlier tests had been misinterpreted and that extensive studies 
demonstrated convincingly that it was safe.^'^ Because of the complexity of the 
scientific questions and the public controversy surrounding cyclamate, the FDA 
Commissioner asked the National Cancer Institute to empanel a group of scientists 
to review and evaluate the data. 

The panel eventually consisted of scientists form the Institute and academic 
consultants. The group met privately over more than a year, reviewed all of the 
data, and in 1976 submitted to FDA a lengthy and, in the Commissioner's view, 
inconclusive report. There had been no dialogue between the NCI panel and the 
Commissioner's office prior to receipt of the report. The panel in substance 
concluded that none of the tests demonstrated that cyclamate was a carcinogen, 
but it went on — in typically cautious language — to describe both the important 
limitations of those studies and the other evidence suggesting that cyclamate 
might pose a chronic health hazard.-^" 

Faced with this guarded judgment, Commissioner Schmidt felt disabled 
with finding that cyclamate would not cause harm to consumers if permitted 
back in food.^'^ Emphasizing that the burden of proving safety lay with the 
petitioner, Abbott Laboratories, the agency denied its petition, precipitating a 
formal administrative hearing and several court challenges. ^'^ The hearing even- 
tually led to a ruling by Schmidt's successor as Commissioner that Abbott had 
failed to carry its burden of proof. -^'"^ 

During the formal administrative process neither party took issue with the 
formation or procedures of the NCI panel. There was, in short, little disagreement 
about the integrity of the process by which the panel's judgment was sought. 
But there was intense dispute over what the panel's judgment should mean for 
FDA. Abbott vigorously challenged its interpretation of specific studies, while 
counsel for FDA's Bureau of Foods went well beyond the panel's report in 
attempting to raise serious doubts about cyclamate 's safety. 

The cyclamate episode suggests that neither the composition nor the pro- 
cedures of the panel influenced the acceptance of its conclusions. Nor is it likely 
that clarification of the panel's charge would have averted the disagreements 
between FDA and the petitioner. The panel members were careful to confine 
their conclusions to the scientific questions about which they felt competent. 
Subsequent dispute stemmed from conflicting views about FDA's legal respon- 
sibility, which led to what was in substance a de novo reexamination of the 
underlying scientific issues. Without expressing any view about the correct res- 

210. Abbott Laboratories" contention and its food additive petition figure prominently in 
Abbott Laboratories v. Harris, 481 P. Supp. 74 (N.D. III. 1979). 

211. The Report of the Temporary Committee for the Review of Data on Carcinogenicity of 
Cyclamate, National Cancer Institute, February 1976, is summarized at 45 Fed. Reg. 61483 (1980). 

212. 41 Fed. Reg. 43754 (1976). § 348 (1976). See generally Merrill, Regulating Carcinogens 
in Food: A Legislator's Guide to the Food Safety Provisions of the Federal Food, Drug, and 
Cosmetic Act, 11 Mich. L. Rev. 171 (1979). 

213. Petition denied, 41 Fed. Reg. 43754 (1976); Hearing granted, 42 Fed. Reg. I25I5 
(1976); Issue Remanded to Administrative Law Judge, 44 Fed. Reg. 47620 (1979). 

214. 45 Fed. Reg. 61473-530 (1980). 


olution of these issues, it is difficult to visualize a procedure that would have 
produced findings immune to relitigation — absent Abbott's agreement in ad- 
vance to abide by the outcome of a particular process. The firm otherwise could 
be expected to contest any conclusions that failed to yield reapproval of cycla- 
mate. Abbott spokespersons would also contend that no peer review procedure 
can be expected to produce unbiased results, so long as FDA retains ultimate 
decisional authority. 

Red No. 40. Red No. 40 is the one remaining red color approved for food 
use in the United States. It was first "listed," i.e., approved, by FDA in 1970 
on the basis of laboratory experiments whose results were sufficiently ambiguous 
to persuade the Canadian government to decline to permit its use.-^'^ A residual 
concern was the possibility that the color might be carcinogenic. In 1975 several 
events conspired to make this possibility a volatile issue for FDA. 

The first event was FDA's decision to suspend its provisional approval for 
Red No. 2, the red color with the longest history of use and widest commercial 
appeal. ^'^ This decision, ironically, was precipitated by a chronic study in FDA's 
own laboratories which, though flawed in execution, indicated that Red No. 2 
might be carcinogenic.^'^ The second event was the receipt of preliminary find- 
ings from an on-going chronic study of Red No. 40, which suggested to some 
reviewers that this color might cause cancer in the test animals. The apparent 
similarity between the two findings made it awkward for FDA to defend its 
action against Red No. 2 and simultaneously resist demands that it ban Red No. 
40. As a legal matter, however, the latter' s approved status made its removal 
more difficult and FDA scientists cautioned that the Red No. 40 findings were 
only preliminary. Furthermore, precipitious action by FDA would have been 
disruptive because the loss of Red No. 2 had prompted most food producers to 
substitute Red No. 40. 

Faced with scientific controversy, FDA created a special "working group" 
of scientists to evaluate the results of the Red No. 40 study as it progressed. ^'^ 
This panel — FDA retained the name "working group" perhaps to defuse claims 
that it was an advisory committee — consisted of scientists from FDA's Bureau 
of Foods, from the Commissioner's own staff, and from parts of the Public 
Health Service. ^'^ The group's responsibility was to monitor the progress of 

215. Red No. 40 is "listed" for use in foods under 21 C.F.R. § 74.340, in drugs under 21 
C.F.R. § 74.1340, and in cosmetics under 21 C.F.R. § 74.2340. 

216. See Certified Color Mfrs. Ass'n v. Mathews, 543 F.2d 284 (D.C. Cir. 1976). 

217. Id. 

218. 41 Fed. Reg. 53546(1976). 

219. Id. The working group was chaired by Dr. Albert C. Kolbye, Jr., Associate Director 
for Science in the Bureau of Foods, FDA, and originally consisted of the following members: Dr. 
Richard Bates, Associate Commissioner for Science. FDA; Dr. Cipriano Cueto, National Cancer 
Institute (NCI); Dr. Morris Cranmer, Director, National Center for Toxicological Research (NCTR); 
Dr. Robert Squire, NCI; Dr. Gary Flamm, NCI; Dr. Charles Frith, NCTR; and Dr. George Wolff, 
NCTR. Members Squire and Frith Later left the Group and were replaced by Dr. Jerrold Ward, 
NCI, and Dr. Morris Weinberger, FDA. In addition. Dr. Kenneth C. Chu, NCI, Dr. John Gart, 
NCI, Dr. David Gaylor, NCTR, and Mr. James Winbush, FDA. all were given permanent positions 
on the Working Group. 42 Fed. Reg. 61630, 61631 (1977). 


ongoing studies of Red No. 40 and report their findings to the Commissioner. 
It periodically met in public session and permitted statements by scientists from 
industry and from the Center for Science in the Public Interest (CSPI), an early 
proponent of banning the color. FDA also released the working group's reports 
for public review, ^^° but it made no formal attempt to assure compliance with 
the FACA. 

In late 1981, FDA Commissioner Hayes released the working group's final 
report, which concludes that Red No. 40 is not carcinogenic in rats or mice.^^' 
The agency accordingly has initiated no regulatory action against the color. It 
now appears unlikely that CSPI will attempt to challenge the group's findings, 
beyond expressing its disagreement and its continued opposition to consumer 
exposure to "untested" food colors. FDA would reject any petition to ban the 
color. CSPI would then have the burden of demonstrating that FDA's failure to 
act was an abuse of discretion^^^ — in the face of a report by expert scientists 
who have been reviewing the accumulating data over nearly six years. 

In addition to time, the creation of the Red No. 40 working group afforded 
FDA not merely the appearance but reality of an independent assessment of the 
color's toxicity. By creating a body to deal specifically with Red No. 40, the 
agency helped to expose its processes for decisionmaking. By combining staff 
members with outside experts, the agency enhanced both its own credibility and 
the standing of its own scientists. Furthermore, the working group's status dem- 
onstrates the sort of attention to the issue that will make FDA's conclusions 
more resiliant to attack than if it rested exclusively on evaluations prepared by 
its own staff. 

Nitrite. Sodium nitrite is added to meat and poultry to prevent spoilage, 
retard the growth of botulinum bacteria, and — most importantly — give meat a 
distinctive and appetizing appearance. Because of evidence that nitrosamines, a 
breakdown product of nitrite, causes cancer in laboratory animals, both FDA 
and USDA have long been concerned about nitrite's safety. ^^-^ In June 1974, 
FDA commissioned Dr. Paul Newberne of MIT to conduct a study to determine 
whether nitrite itself is carcinogenic. Four years later Dr. Newberne submitted 
his finding that nitrite appeared to cause cancer in rodents. 

When this finding was reported to FDA in 1978, the agency had no formal 
written procedure for evaluating data generated by contract studies, but its prac- 
tice was well established. '^'^'^ Bureau of Foods scientists would usually conduct 
in-house evaluations to assess the strengths and weaknesses of the data and 

220. See, e.g. 42 Fed. Reg. 8005 (1977); 44 Fed. Reg. 30437 (1979). 

221. 46 Fed. Reg. 51037 (1981). 

222. Section 706(2) (A) of the Administrative Procedure Act. 5 U.S.C. § 706(a)(A), provides 
that a reviewing court shall set aside agency action, findings, and conclusions found to be "arbitrary, 
capricious, an abuse of discretion, or otherwise not in accordance with law." 

223. See General Accounting Office. Does Nitrite Cause Cancer? Concerns About Validity 
of FDA-Sponsored Study Delay Answer (1980) at 9- 12 [hereafter GAO Nitrite Report]. 

224. Id. at 12. 


explore regulatory alternatives. ^^^ The Bureau's Division of Toxicology generally 
would commence the review and, if cancer were suspected, forward the study 
to the Bureau's Cancer Review Committee. That committee could recommend 
establishment of an interagency working group, similar to that created for Red 
No. 40, if the compound studied posed difficult scientific or regulatory issues. 
This internal review process did not ordinarily include reexamination of tissue 
slides to confirm reported pathological findings. ^^^ 

In the case of the nitrite study, FDA omitted routine review by the Bureau 
of Foods; instead. Commissioner Kennedy set up a special task force of selected 
agency personnel to evaluate Newberne's results. -^^^ According to FDA officials 
this departure from routine was a response to the special problems posed by 
nitrite: It is added to seven per cent of U.S. food, which represents 12 billion 
dollars per year in sales, and its use prevents fatal poisonings. Newberne's 
reported findings were thus thought to demand immediate attention in a setting 
that permitted simultaneous, top-level evaluation of regulatory options. The 
Commissioner's special task force was accordingly directed to consider the sci- 
entific, legal, and regulatory dimensions of the nitrite problem. After the task 
force's one toxicologist confirmed Newberne's reported findings, the members 
devised a plan for a gradual phase out of nitrite, which was to be disclosed at 
the same time the MIT data were released publicly. ^^^ 

Before FDA and USDA announced Newberne's results. Bureau of Foods 
scientists raised several potential problems with the study. ^^^ These problems 
were sufficiently serious that the Bureau's Acting Director recommended to 
Commissioner Kennedy that a special working group of scientists should be 
formed to evaluate the Newbeme study's validity. Accordingly, on August 8, 
1978, the agency created the Interagency Working Group on Nitrite (lAWG). 
This group included scientists from FDA, USDA, NIC, and the NIEHS, who 
were to evaluate the accuracy of Newberne's pathological diagnoses. 

After two lAWG pathologists reexamined a sample of Newberne's slides 
and failed to confirm his diagnoses, the working group recommended thorough 
independent reexamination of the study records. Based on this recommendation, 
FDA contracted with Universities Associated for Research and Education in 
Pathology, Inc. (UAREP), a non-profit consortium of pathologists from 15 uni- 
versities, to review Newberne's pathology. The working group later reported 
that both it and UAREP had concluded that the evidence did not support New- 

225. /^. at 9- 16. 

226. FDA will undertake to verify pathology results when they seem internally inconsistent, 
when the pathology descriptions are ambiguous or do not agree with the conclusions of the study, 
or when the results do not coincide with available data. 

227 . The task force included the Commissioner, the Deputy Commissioner, the Chief Counsel , 
the Special Assistant to the Commissioner, the Executive Assistant to the Commissioner, the Acting 
Director, Bureau of Foods, the Project Officer for the MIT nitrite contract and a toxicologist from 
the Division of Toxicology, Bureau of Foods. 

228 . G AO Nitrite Report at 14-21. 

229. /rf. at 21-22. 


heme's findings that nitrite caused cancer in rodents. Red-faced, FDA and USDA 
simultaneously announced that no action to end the use of nitrite in food would 
be initiated. ^''^^ 

The nitrite episode demonstrates both the importance and the difficulty of 
peer review of laboratory findings. FDA's missteps stemmed from tension be- 
tween its role as public advisor, which could have been undermined by the 
release of the Newberne findings before it had devised a response, and its 
responsibility to rely on sound science. The agency had only recently weathered 
the public furor over saccharin. It therefore is understandable why, when they 
received Newberne 's findings, agency officials wanted to find alternatives to a 
complete ban. But its preoccupation with the regulatory implications of New- 
berne 's study led the Commissioner's task force to subordinate its scientific role. 
Only after the task force recommended a phase out of nitrite did FDA begin to 
explore the questions about the study's validity, questions raised initially by staff 
scientists who had been excluded from the special review procedure set up for 
the occasion. The agency's subsequent embarrassment indicates the importance 
of scientific peer review, but the episode illustrates the capability of FDA's own 

b. Consumer Product Safety Commission. Of the four agencies studied, 
the CPSC has had least experience with advisory panel review of chemical 
hazards. It is perhaps ironic, therefore, that the Commission is now obligated 
by statute to consult with an independent scientific review panel before attempting 
to regulate products believed to present a risk of cancer. This obligation is the 
product of congressional amendments to the CPS Act (and other statutes ad- 
ministered by the Commission) enacted in 1981.^^' 

The CPSC's past failure to consult with independent scientists has not 
betokened indifference or special confidence in its own scientific staff. With the 
exception of TRIS, all of the substances that the agency attempted to regulate 
during the 1970's had previously been characterized as carcinogens by other 
scientific bodies or sister regulatory agencies. ^^^ Thus its findings that vinyl 
chloride, asbestos, benzene, and chloroflurocarbon propellants posed a potential 
risk of cancer were never seriously challenged. ^"^"^ It is not surprising, therefore, 
that the Commission saw no need for independent peer review. The agency's 
conclusions that products containing these ingredients posed a significant risk, 
however, often did provoke disagreement because of uncertainty about the in- 
tensity, duration, and route of human exposure. The agency never sought in- 
dependent review of its exposure assessments before initiating regulation. ^^"^ 

230. Chemical Regulation Reporter (BNA), 663-64 (1980). 

231. Consumer Product Safety Amendments of 1981, Pub. L. No. 97-35, § 1206, 95 Stat. 
357 (1981) (amending 15 U.S.C. § 2077 (1976)). 

232. Merrill, CPSC Regulation of Cancer Risks in Consumer Products: 1972-1981, 67 Va. 
L. Rev. 1261, 1310 (1981) [hereafter Merrill: Consumer Products]. 

233. Id. at 1310. 

234. Uncertainty about exposure was troublesome in the case of urea formaldehyde foam and 
hair dryers containing asbestos. Id. at 1346-51, 1354-60. 



Formaldehyde. In 1980, the CPSC for the first time formally sought the 
advice of outside scientists in evaluating a chemical's risks. When the agency 
contemplated regulating urea formaldehyde foam insulation (UF insulation) it 
confronted a chemical whose carcinogenicity had not been confirmed by scientific 
review or established in regulatory actions by other agencies. 

UF foam is a cellular plastic product with a shaving cream-like consistency 
which is pumped into the walls of homes through small holes to provide thermal 
insulation. In 1976 the CPSC learned about residents of homes insulated with 
UF foam who had begun to complain of respiratory distress, persistent cough, 
nausea, dizziness, and other irritations, which some scientists attributed to es- 
caping formaldehyde gas.^-^^ The CPSC proposed a warning requirement for UF 
insulated homes, based on these acute effects. ^^^ When the Chemical Industry 
Institute for Toxicology (CIIT)^"^^ in late 1970 reported preliminary results of an 
inhalation study which showed formaldehyde to be a carcinogen in rodents, the 
agency concluded that it should consider more drastic action. 

Accordingly, on February 12, 1980 the Commission announced the for- 
mation of a panel composed of scientific experts from other regulatory and 
research agencies to review the data on formaldehyde (including specifically the 
CUT study) and assess the cancer risk of formaldehyde gas leaking from UF 
insulation. ^^^ The review panel was established by the National Toxicology 
Program at the request of the CPSC and the other IRLG agencies. The CPSC 
offered no detailed explanation for its decision to have a panel of outside experts 
assess the risks of UF foam insulation, but its reasoning is easily deduced. The 
Commissioners were surely cognizant of the limitations of the agency's own 
small scientific staff. More important, the CPSC had frequently made clear its 
desire to maintain a common front with the other IRLG agencies, which had 
placed formaldehyde on collective priority list for joint evaluation and regula- 
tion. ^^^ The interagency panel's findings were expected to guide subsequent 
decisions of the other three agencies as well as the CPSC.'^'*^ 

The review panel consisted exclusively of government scientists, most from 
research agencies such as NIEHS, NCTR, and NCI.^^' FDA, EPA, and OSHA 
also supplied members, and the secretary was Dr. Peter Preuss, CPSC's Associate 
Executive Director for Health Sciences. The panel met in closed session on 

235. See id; Troxel, Interagency Coordination of Toxic Substance Regulation: The Formal- 
dehyde Controversy, unpublished essay. University of Virginia School of Law (Dec. 21, 1981). see 
also 45 Fed. Reg. 34031, 39434, 39438 (1980); 46 Fed. Reg. 11188. 11190 (1981). 

236. See 45 Fed. Reg. 39434 (1980). 

237. The CUT is a scientific organization supported by thirty-six American chemical cor- 
porations and located in Research Triangle Park, North Carolina. It tests chemicals to determine 
their health effects and seeks to develop and improve test protocols. The CUT enjoys a reputation 
for high-quality scientific work. 

238. See 46 Fed. Reg. 1 1 190 (1981). 

239. Merrill Consumer Products at 1262, 1267. 

240. Interview with Dr. Peter Preuss, note 163, supra. 

241. 46 Fed. Reg. 11190(1981). 


several occasions to review the CUT findings and other data on formaldehyde 
toxicity. The panel also consulted with the CUT scientists responsible for the 
inhalation study and accepted oral as well as written submissions from the 
Formaldehyude Institute, an association of users and manufacturers. The panel 
submitted a written report to the CPSC in late November 1980, somewhat later 
than expected. ^"^^ Based on evaluation of the CUT study results and other data, 
the panel members were convinced that formaldehyde was an animal carcinogen 
and should be considered a potential human carcinogen. The panel did not attempt 
to estimate the risk posed by exposure to formaldehyde fumes in UF foam 
insulated homes. The Commission's staff later performed such a risk assessment, 
using the panel's recommended extrapolation procedure and its own estimates 
of consumer exposure conditions. ^"^^ 

It is premature to assess the impact or utility of the CPSC's initial experiment 
with independent peer review. Based on the panel's assessment, the agency later 
proposed a ban on future installation of UF foam insulation. ^"^"^ This proposal 
drew sharp opposition and sharp criticism from the industry of both the panel's 
and the agency's characterization of the risks of formaldehyde.^'*'' After once 
postponing a decision on its proposal, ^"^^ the Commission in late February, 1982, 
voted to ban further installation of UF foam insulation. ^"^^ The announced ra- 
tionale for this action departs somewhat from the agency's proposal, which 
stressed the cancer risk of escaping formaldehyde gas; the final ban is justified 
primarily in terms of the acute health effects of formaldehyde gas, which were 
the basis of the agency's original proposal to require a warning on UF insulated 
homes. ^"^^ 

The CPSC's reemphasis of the acute hazards of formaldehyde position does 
not necessarily betray second thoughts about its carcinogenicity, for which the 
animal evidence now seems unequivocal,-^*^ or about the advice of the expert 
review panel. Yet the Commission or its agency staff clearly have become 
reluctant to make the risk of cancer the central ground for a ban of UF foam 
insulation. Undoubtedly contributing to this uncertainty was EPA's near-con- 
temporaneous, and widely predicted, announcement that, given prevailing ex- 
posure levels, the evidence of formaldehyde's carcinogenicity did not justify 
regulation under section 4(f) of the Toxic Substances Control Act.'^''^ Previously, 
OSHA had declined to issue an emergency temporary standard for worker ex- 

242. The report was delivered on November 21, 1980. 46 Fed. Reg. 1 1 188, 11 190 (1980). 

243. Id. at 11188, 11190-92, 11195. 

244. 46 Fed. Reg. 11188 (1981). 

245. [1981] Chemical Regulation Reporter 1209 (BNA). 

246. 46 Fed. Reg. 49140 (1981). 

247. [1982] Chemical Regulation Reporter (BNA). 

248. Id. at 33. 

249. Interview with Dr. Norton Nelson. New York University, March 18, 1982. 

250. TSCA § 4(f), 15 U.S.C. § 2603 (4)(0. 


posures to formaldehyde, though it reportedly contemplates rapid action to es- 
tablish a permanent health standard to reduce worker exposure. ^^^ 

198] Amendments to CPS Act. It may not be possible to assess the impact 
of advisory panel review on the CPSC's regulation of chronic hazards, but such 
review will hereafter be an integral part of the agency's decisionmaking. Early 
in 1981, the Reagan Administration invited Congress to consider abolishing the 
CPSC and transferring its functions to the Department of Commerce. ^^^ While 
Congress declined to disband the agency, it reauthorized it for only two years 
and reduced its budget by 30 percent. ^^"^ The reauthorizing legislation also made 
important changes in the CPSC's authority and procedures. Among the changes 
in the CPS Act is a new section 31, which directs the agency to seek expert 
advice whenever it contemplates regulation of a substance believed to pose a 
risk of cancer. ^^"^ 

251. [1981] Chemical Regulation Reporter 1 170 (BNA). 

252. Merrill, Consumer Products at 1373-75. 

253. Id. at 1373, n.583. 

254. Amended CPS Act § 31, 15 U.S.C. § 2080, reads in pertinent part as follows: 

(b)(1) The Commission may not issue an advance notice of proposed rulemaking for — 

(A) a consumer product safety rule, 

(B) a rule under section 27(e), or 

(C) a regulation under section 2 (q)(l) of the Federal Hazardous Substances Act, 
relating to a risk of cancer, birth defects, or gene mutations from a consumer product 
unless a Chronic Hazard Advisory Panel, established under section 28, has, in accordance 
with paragraph (2), submitted a report to the Commission with respect to whether a 
substance contained in such product is a carcinogen, mutagen, or teratogen. 

(2)(A) Before the Commission issues an advance notice of proposed rulemaking for — 

(i) a consumer product safety rule 

(ii) a rule under section 27(e), or 

(iii) a regulation under section 2(q)(l) of the Federal Hazardous Substances Act, 
relating to a risk of cancer, birth defects, or gene mutations from a consumer product, the 
Commission shall request the Panel to review the scientific data and other relevant infor- 
mation relating to such risk to determine if any substance in the product is a carcinogen, 
mutagen, or a teratogen and to report its determination to the Commission. 

(B) When the Commission appoints a Panel, the Panel shall convene within 30 days 
after the date the final appointment is made to the Panel. The Panel shall report its 
determination to the Commission not later than 120 days after the date the Panel is convened 
or, if the Panel requests additional time, within a time period specified by the Commission. 
If the determination reported to the Commission states that a substance in a product is a 
carcinogen, mutagen, or a teratogen, the Panel shall include in its report an estimate, if 
such an estimate is feasible, of the probable harm to human health that will result from 
exposure to the substance. 

(C) A Panel appointed under section 28 shall terminate when it has submitted its report 
unless the commission extends the existence of the Panel. 

(D) The Federal Advisory Committee Act shall not apply with respect to any Panel 
established under this section. 

(c) Each Panel's report shall contain a complete statement of the basis for the Panel's 
determination. The Commission shall consider the report of the Panel and incorporate such 
report into the advance notice of proposed rulemaking and final rule. 


Specifically, new section 31 provides that the CPSC may not issue an 
advanced notice of proposed rulemakings''^ "relating to a risk of cancer," unless 
it has first recieved a report from a "chronic hazard advisory panel. "-^^^ Such 
a panel is a group of seven scientific experts appointed by the Commission, 
from nominees submitted by the National Academy of Sciences, to review the 
scientific data and "determine if [a] substance ... is a carcinogen." If feasible, 
the panel shall also provide an estimate "of the probable harm to human health 
that will result from exposure to the substance. "^^^ A panel does not become a 
permanent part of the Commission's structure but is created to evaluate a specific 
hazard. Panel members must have "demonstrated the ability critically to assess 
chronic hazards to risks to human health" from toxic substances; none may be 
employees of the federal government. ^^^ 

The amended CPS Act does not describe the procedures by which chronic 
hazard advisory panels shall perform their function. The statute does state that 
a panel may request information from the states, industry, and other private 
sources, as well as from the Commission, but it may obtain or release information 
only through the agency. ^^^^ A panel must submit a written report, containing 
"a complete statement of the basis for [its] . . . determination, "^^^ which shall 
be incorporated into any ANPR that the agency decides to issue but is not to 
bind the agency. The statute also provides that "the Federal Advisory Committee 
Act shall not apply with respect to any Panel established under this section, "-s^' 
thus apparently freeing panels from the obligations to provide advance notice of 
meetings and to open their deliberations to the public. ^^^ 

The scant legislative history of the 1981 CPSC amendments reveals little 
about Congress' understanding of these requirements for peer review, but their 
enactment indicates skepticism of the CPSC's ability to assess chronic hazards. ^^^ 
The authors of the provision may or may not have considered the formaldehyde 
panel a worthwhile experiment, but the exclusion of federal scientists reveals a 
desire to remove the peer review function from government altogether. It has 
been suggested that section 31 represents the first attempt to incorporate into 
federal law the "national science council" concept advocated by Congressman 
William Wampler.^^ 

c. Occupational Safety and Health Administration. The OSH Act contem- 
plates two types of advisory committees to assist OSHA.^^^ The Act itself 

255. Id. 

256. Id. 

257. Id. 

258. CPS Act § 28(b)(2), 15 U.S.C. § 2077 (b)(2) (as amended 1981). 

259. Id. at § 28(g), (h)(1), 15 U.S.C. § 2077(g), (h)(1) (as amended 1981). 

260. CPS Act § 31(b)(2KB), (C), 15 U.S.C. § 2080(b)(2)(B), (C) (as amended 1981). 

261. Id. at § 31(b)(2)(D), 15 U.S.C. § 2080(b)(2)(D) (as amended 1981). 

262. See notes 173-91 supra and accompanying text. 

263. House Conf. Rep. No. 97-208, 97th Cong., 1st Sess. 881-83 (1981). 

264. See Risk Assessment in the Federal Government, p. 135-139. 

265. OSH A Draft at 14. 



establishes a standing body, the National Advisory Committee on Occupational 
Safety and Health (NACOSH),^^^ and it also empowers OSHA to create special 
expert committees to provide advice on specific standards. ^^^ Both NACOSH 
and ad hoc committees have been involved in OSHA's efforts to set workplace 
health standards, but their role has always been a secondary one and over time 
has declined in importance. 

NACOSH has approximately twenty members who are drawn from the 
industrial health community, from industry, and from employee organizations. ^^^ 
These members are not expected to be expert in evaluating specific hazards but 
rather are viewed as representative of the communities primarily interested in 
occupational health and safety. NACOSH meets on a regular schedule to provide 
advice on broad policy issues, e.g., OSHA enforcement practices or the allocation 
of agency resources. ^^^ While the Committee generally encouraged OSHA's 
mid-1970's shift of focus from safety to health hazards, it has played no sig- 
nificant role in the development of specific OSHA standards or of OSHA's 
generic carcinogen policy. NACOSH's broader influence, moreover, has always 
depended on the receptivity of the incumbent OSHA administrator. Before Dr. 
Eula Bingham became Assistant Secretary, she viewed the advisory committee 
process as an important means of exposing the agency to new ideas and good 
science. Afterwards, she came to regard committees as a cause of delay and a 
hindrance to effective decisionmaking. She attended only the first and last meet- 
ings of NACOSH during her tenure. ^^^ 

OSHA has also exercised its authority to appoint special advisory com- 
mittees in developing specific workplace standards. Membership on such ad hoc 
committees is limited to fifteen or fewer members, at least one of whom must 
be appointed by the Secretary of HHS.^^' The membership of these committees, 
too, must be balanced to assure representation of the views of employers, work- 
ers, and state public health and safety agencies. "^^ Labor organizations, and to 
a lesser extent employers as well, have viewed these ad hoc committees as 
vehicles for making their views known to the agency, and have pressed to be 
represented among the membership. ^^"^ 

The partisan outlook of many members of these ostensibly specialized ad 
hoc committees has not always been disabling, perhaps because the committees 
have typically been consulted on the full range of policy issues OSHA confronts 
in setting health standards, not merely, or even primarily, health effects ques- 
tions. For example, in 1973 the steel industry petitioned OSHA to raise the then- 

266. Id. 

267. Id. 

268. 29 U.S.C. § 656(b). Interview with Charles Gordon, note 164 supra. 

269. Interview with Charles Gordon, note 164 supra; Interview with Grover Wrenn, former 
Director of Health Standards, OSHA, June 23, 1981. 

270. Interview with Grover Wrenn, note 269 supra. 

271. 29 U.S.C. § 656(b). 

272. Id. 

273. Interview with Grover Wrenn, note 269 supra. 


existing exposure limits for coke oven emissions. ^^^ OSHA Administrator Morton 
Corn established a special advisory committee with Eula Bingham, then still an 
academic, as chair. After reviewing available data, the committee recommended 
stricter exposure limits than those prevailing. Industry members had provided 
most of the data supporting the feasibility of the recommended standards, and 
some privately agreed with the majority recommendation. ^^^ Nonetheless, con- 
cern for their constituents prevented them from supporting OSHA's eventual 
proposal to amend the coke oven standard. The reluctance of many members of 
ad hoc committees to express views unpopular with the groups they represent 
has led many OSHA officials to question Congress' original conception. ^^^ 

In its early years, OSHA routinely appointed an advisory committee for 
each health hazard it contemplated regulating, ^^^ but eventually abandoned the 
practice. The committee established to review the coke oven standard was the 
last such body that OSHA has consulted in setting a health standard. Dr. Eula 
Bingham, who became Assistant Secretary in 1977, felt competent to assess 
most scientific data personally; and agency staff found that committee consul- 
tation delayed the standard-setting process without contributing useful advice or 
insights. ^^^ Perhaps equally important, the agency found it possible to engage 
expert scientists as consultants without going through the process of establishing 
formal committees. Few observers of OSHA's performance during the last decade 
have applauded the contribution of ad hoc advisory committees or criticized the 
agency's failure to create new ones. No committee's endorsement ever dimin- 
ished criticism of OSHA's scientific analysis; every recent OSHA health standard 
has been vigorously contested at the administrative level, and all but two have 
been challenged in court. ^^^ 

OSHA's dissatisfaction with advisory committees does not, however, dem- 
onstrate that scientific peer review cannot be successful in this context. The 
American Industrial Health Council's proposal for a "science council," dis- 
cussed in more detail in the next Section, ^^^ was inspired primarily by experience 
with OSHA's health standards process. Furthermore, as noted, the agency's own 
neglect of the advisory committee mechanism did not display a desire to dispense 
with outside scientific advice; the agency often obtained such help by appointing 
consultants. It is at least plausible, therefore, that a mechanism for assuring 
independent consultation on scientific issues would improve OSHA's decisions 
if it did not also have to fulfill the political objective of assuring representation 
to both employer and worker interests. 

274. See 38 Fed. Reg. 26207 (1973). 

275. Interview with Grover Wrenn, note 269 supra. 

276. Id. H 260. See Nichols and Zeckhauser, OSHA Comes to the Workplace, 54 Pub. Int. 

277. Interview with Charles Gordon, note 164 supra. 

278. Interview with Grover Wrenn, note 269 supra. 

279. See OSHA Draft at 39. 

280. See text accompanying notes 341-59 infra. 


d. Environmental Protection Agency. EPA's experience with expert ad- 
visory panels is distinctive. It is the only agency among the four studied whose 
risk assessments — in this case for individual pesticides — have for some time 
been subject to mandatory peer review. The mechanism for this review is the 
Scientific Advisory Panel (SAP), a standing committee mandated by statute. 

SAP was established by Congress in 1975 to strengthen EPA's scientific 
analysis. ^^' From the outset the panel's principal responsibility has been to 
provide impartial oversight of EPA's own evaluation of the environmental and 
health risks posed by particular pesticide uses. SAP's creation reflected sentiment 
in Congress that previous EPA actions against pesticides had lacked an inde- 
pendent scientific base."^^~ The legislative history of the 1975 amendments to 
FIFRA reveals a desire to introduce disinterested scientific review into EPA's 
regular decisionmaking process: 

The creation of a Scientific Advisory Panel and notification of this 
panel of proposed cancellation or suspension actions of any proposed reg- 
ulations are intended to further assure balance and objectivity in EPA ac- 
tions. The purpose of this provision is to assure that the EPA obtains 
unbiased objective scientific opinion in making its decisions. ^^^^ 

SAP's responsibilities have been enlarged by Congress on two subsequent 
occasions. As amended in 1975, FIFRA required only that SAP be consulted 
when the agency contemplated cancellation of a pesticide's registration. ^^"^ In 
1978 Congress also required that the EPA Administrator solicit SAP's recom- 
mendations for operating guidelines to improve the quality of scientific analyses 
performed by agency personnel to support agency decisions. ^^'^ Two years later, 
Congress again enlarged SAP's responsibilities. First, it obligated EPA to afford 
the panel an opportunity to evaluate all suspension actions "as to the impact on 
health and the environment. "^^^ Second, SAP — or other nonagency scientists 
selected from a list it assembled — was directed to review the "design, protocols, 
and conduct of major scientific studies" relied upon by the agency in changes 
of classification, suspension, or cancellation proceedings. ^^^ The panel also was 

281. Pub. L. No. 94-140, § 7, 89 Stat. 753-54 (1975), codified at 7 U.S.C. §§ 136d, 136w 
(1976). EPA Administrator Train opposed SAP as "unnecessary." S. Rep. No. 94-452, 94th Cong., 
1st Sess. 19, reprinted in [1975] U.S. Code Cong. & Ad. News 1375 (statement of Russell E. 

282. See, e.g., S. Rep. No. 94-452, 94th Cong., ist Sess. 9, reprinted in [1975] U.S. Code 
Cong. & Ad. News 1366; H. Conf. Rep. No. 94-668, 94th Cong., 1st Sess. 5, reprinted in [1975] 
U.S. Code Cong. & Ad. News 1380. 

283. S. Rep. No. 94-452, 94th Cong., 1st Sess. 9, reprinted in [1975] U.S. Code Cong. & 
Ad. News 1366. 

284. Pub. L. No. 95-396, § 23, 92 Stat. 836 (1978), codified at 7 U.S.C. 136w (Supp. Ill 

285. Pub. L. No. 96-539, §§ 1, 2(a), 94 Stat. 3194, 3195 (1980), codified at 7 U.S.C. § 
I36w (Supp. IV 1980). 

286. Id. 

287. H.R. Rep. No. 96-1020, 96th Cong., 2d Sess. 4, 18, reprinted in [1980] U.S. Code 
Cong. & Ad. News 6723, 6733. 


given power to appoint additional subpanels of outside experts to assess the 
studies relied upon in agency proposals. These most recent changes were a 
product of Congressional dissatisfaction with EPA's suspension of certain uses 
of the herbicide 2,4,5-T.^^^ They were also a response to a General Accounting 
Office study which criticized EPA for failure to develop a formalized system of 
peer review. ^^^ 

SAP's role and structure. Throughout its brief history, SAP has thus been 
conceived of as a mechanism for overseeing the scientific foundations of EPA's 
pesticide decisions. By law the panel must be afforded an opportunity to evaluate 
notices of intent to cancel or reclassify pesticide uses,^^^ contemplated suspen- 
sions of registration, and proposed and final general pesticide regulations. Such 
agency decisions are to be submitted to SAP "for comment as to the impact on 
health and the environment of the action proposed. "^^' In no instance, however, 
are the panel's recommendations binding on EPA. 

SAP consists of seven members selected by the EPA Administrator from 
among six nominees by NIH and six nominees submitted by the National Science 
Foundation. ^^^ As of July, 1981, SAP had only six members, whose fields 
included plant pathology, epidemiology, zoology- wildlife, pharmacology, en- 
tomology, biochemistry, and pathology. ^^-^ For several months the panel had no 
member with substantial background in toxicology, a critical speciality in as- 
sessing the health risks posed by pesticides. According to some observers, lack 
of such expertise has been a persistent problem during much of SAP's exis- 
tence. ^^"^ 

From its creation SAP has generally met once a month. ^"^^ The panel's 
meetings are routinely open to the public, and interested parties are generally 
encouraged to make presentations. ^^^ Indeed, SAP has sometimes allowed ses- 
sions to become adversarial debates on issues that can range well beyond the 
health and environmental risk assessments contained in the documents submitted 
for its formal review. ^^^ The panel may request Office of Pesticide Programs 
representatives to attend such meetings to defend agency determinations and in 
the same session permit industry representatives and other interested persons to 
respond. In the process individual panel members may actively question partic- 

288.- Id. at 4, [1980] U.S. Code Cong. & Ad. News 6723. 

289. Id. 

290. FIFRA § 25(d). 7 U.S.C. § 136w (Supp. IV 1980). 

291. Id. 

292. Id. 

293. Membership List. Scientific Advisory Panel. EPA, as of July 13, 1981. 

294. Interview with Philip Gray, Executive Secretary, Scientific Advisory Panel, EPA (July 
14, 1981). 

295. Id. 

296. Id. 

297. Interview with Steven Jellinek, former Assistant Administrator for Pesticides and Toxic 
Substances, EPA (June 23, 1981). 


There has been concern about SAP's willingness to permit rigorous debate 
among public participants. ^^^ Some EPA officials have feared, for example, that 
pesticide manufacturers might be able to mount flashy, well-financed presen- 
tations before SAP that would undermine crucial agency positions after the 
expenditure of considerable agency resources. In 1980 Congressman Fithian 
expressed concern that the public character of SAP meetings — required by the 
FACA rather than EPA policy — had made the panel's proceedings more ad- 
versarial and less advisory than originally contemplated: 

If we have had some other accretions along the way so that it has 
gotten sort of like an administrative law judge proceedings where you bring 
in both sides and you argue the case, then I would submit that it is a 
perversion of the intent of the Scientific Advisory Panel which was never 
intended to do that. 

It was purely and intentionally to go to the best scientists we could 
command and say — for we are not scientists on this committee — "Are we 
using the best scientific data in making this decision on the protection of 
the environment and yet allowing the widest possible use of safe chemi- 
cals?" That was the purpose behind it.^^^ 

Apprehensions about SAP's vulnerability, however, have not been borne 
out by experience. The June, 1981 meeting of SAP to consider EPA's proposed 
actions on wood preservatives is a useful illustration. During the three-day session 
the American Wood Preservers Institute (AWPI) launched an uncompromising 
attack on the scientific underpinnings of EPA's decision to restrict the permitted 
wood uses of pentachlorophenol, inorganic arsenicals, and creosote. ^*^^ The AW- 
PI' s apparent objective was to discredit OPP's proposals by refuting all evidence 
of adverse health effects caused by the suspect compounds. The industry's strat- 
egy backfired. SAP endorsed virtually all of EPA's findings and recommenda- 
tions. ^°' Furthermore, its draft summary of the proceedings included this 
uncharacteristically critical language: 

The Panel wishes to express its appreciation to Special Pesticide Review 
Division and Hazard Evaluation staff for their informative briefings. . . . 
On the other hand, the Panel deplores the lack of scientific objectivity of 
the presentations by industry concerning biological hazards of wood pre- 
servatives, and finds the industry's denial of scientific data concerning the 

mutagenicity and carcinogenicity of the wood preservatives to be disturb- 

298. Id. 

299. Proposed Extension of the Federal Insecticide, Fungicide, and Rodenticide Act: Hearings 
on H.R. 70 J 8 Before the Subcomm. on Department Investigations, Oversight, and Research of the 
House Comm . on Agriculture , 96th Cong . . 2d Sess . 185(1 980) (Comments of Congressman Fithian) . 

300. Interview with William Dickinson, Deputy Director, Special Pesticide Review Division, 
Office of Pesticide Programs, EPA (July 13. 1981). 

301. Interview with William Dickinson, supra note 300, interview with Philip Gray, supra 
note 294. 

302. Scientific Advisory Panel, EPA, Review of Preliminary Notice of Determination Con- 
cluding the Rebuttable Presumption Against Registration (RPAR) on Wood Preservative Uses of 
Pentachlorophenol, Inorganic Arsenicals, and Creosote (May 15, 1981) (draft). 


This episode illustrates the distinction between SAP's statutory review func- 
tion and formal cancellation hearings. Perceptive participants before the panel 
have generally stressed the dominance of benefits over risks rather than criticizing 
the competence of agency scientists. ^^-^ Some participants, however, appear 
simply to be making a record before SAP that might be useful in later proceedings, 
rather than addressing new evidence to the panel or enlarging them in discussion 
of the underlying scientific issues. ^^"^ 

Record of SAP review. Critics of EPA's early cancellation actions may have 
expected the SAP would repudiate the agency's scientific assumptions and reject 
many of its proposals, but this has not occurred. Over the past five years SAP 
has agreed with the great majority of OPP judgments brought before it. One 
informed estimate has placed SAP's approval of staff risk assessments at as high 
as 90 percent, ^^^ and other agency officials concur that SAP has affirmed major 
agency decisions "almost without fail."-^^^ 

The paradigm but perhaps unique case of disagreement between the panel 
and EPA grew out of the agency's handling of 2,4,5-T. After concluding that 
forest, pasture, and right-of-way applications of the herbicide posed an imminent 
threat to the public health, EPA in March, 1979 issued an emergency suspension 
order and a notice of intent to cancel these three uses of 2,4,5-T and its companion 
product, Silvex.^^^ FIFRA did not then require SAP oversight of EPA suspension 
decisions, but the agency, as it was required to do, later sought the panel's 
comments on its intention to hold a hearing on the appropriate regulatory dis- 
position of the other uses of 2,4,5-T and Silvex.''^^ After three days of public 
meetings, SAP declined to support the agency's proposed actions. The panel 
believed that additional data, including further testing for oncogenicity and re- 
productive toxicity and more complete monitoring for residues, were required 
before a hearing could be profitably held.-^^*^ Despite SAP's disapproval EPA 
held firm, announcing the commencement of broad ranging hearings on the risks 
and benefits of 2,4,5-T, which began in March of 1980. "^'^ Two major conse- 
quences flowed from this series of events: Protracted hearings over 2,4,5-T 
between EPA and Dow Chemical, and ongoing settlement negotiations, have 
yet to be concluded. And congressional displeasure with EPA's handling of 
2,4,5-T/Silvex crystallized in the 1980 amendment mandating SAP's involve- 
ment in future suspension decisions.^" 

303. Interview with Philip Gray, supra note 294. 

304. Id. 

305. Interview with Steven Jellinek, supra note 297. 

306. Interview with Philip Gray, supra note 294. 

307. 44 Fed. Reg. 15874 (1979). 

308. Memorandum from Dr. H. Wade Fowler, Jr., Executive Secretary, SAP. to Deputy 
Assistant Administrator for Pesticide Programs, Review of FIFRA Section 6(b)(2) Action on 2,4,5- 
T and Silvex (Sept. 27, 1979). 

309. Id. 

310. Id. 

311. See authorities cited note 282 supra. 


In April 1979 two members of SAP, Vice Chairman Robert A. Neal and 
John E. Davies, appeared at congressional oversight hearings on EPA's suspen- 
sion of 2,4,5-T/Silvex and DBCP.'^'^ Neal expressed concern that the SAP had 
been bypassed in both instances, and his recommendation that SAP be formally 
involved in suspension decisions was implemented in 1980. But neither Neal 
nor Davies agreed with suggestions that there had been friction between EPA 
and SAP. Neal at one point complimented the agency for ''doing a reasonably 
good job."^'^ 

EPA's conception of SAP's role was later explained before the same sub- 
committee by Steven Jellinek, then Assistant Administrator for Pesticides and 
Toxic Substances: 

In summary, sometimes we do disagree with the SAP and find that a 
more or less stringent measure than the one they support is most appropriate, 
considering the benefits as well as the risks of the pesticide uses in question. 
Such differences are not surprising, however, considering that risk assess- 
ment is not an exact science, but rather a process which, while scientifically 
supported and necessary for regulatory action, is still full of uncertainties. 
These differences are usually administrative differences rather than scientific 
differences. We continue to respect and value the SAP's viewpoints and 
consider the Panel as an important contributor to balanced decisionmak- 

Assessment. It is the consensus of present and former EPA officials that 
SAP has played a valuable role in the pesticide program.^ '^ The agency has been 
able to attract the panel members who are highly regarded within their own 
specialities and who are willing to serve, with only modest compensation, be- 
cause they believe that the scientific issues underlying pesticide regulation do 
in fact influence decisions that have major economic and social impact. -^'^ More 
than once SAP has defused criticism of EPA by endorsing an agency recom- 
mendation or confirming the agency's evaluation of a particular compound. ^'^ 
Because its members are respected, SAP enjoys cautious support from pesticide 
and chemical manufacturers. One industry spokesman has even remarked that 
"SAP represents the highest degree of technical expertise and competence in 
the scientific community, and has functioned as a model of objectivity."^'^ 

SAP's role has also improved internal EPA decisionmaking. Mandatory 
review of notices of intent and proposed regulations has encouraged the OPP to 
prepare more thoroughly. ^'^ In addition, OPP's Hazard Evaluation Division has 

312. Hearings Before the Subcomm. on Department Investigations, Oversight, and Research 
of the House Comm. on Agriculture, 96th Cong., 2d Sess. (1979). 

313. Id. 

314. Id. 

315. Interview with Steven Jellinek, supra note 297; interview with Philip Gray, supra note 

316. Interview with Steven Jellinek, supra note 297. 

317. Id. 

318. Hearings, supra note 299, at 24. 

319. Interviews with members of the Office of General Counsel, EPA (June 23, 1981). 


consulted more frequently with panel members to arrive at findings that will 
withstand SAP's normally skeptical analysis. The agency's Exposure Assessment 
Group was created in part to strengthen presentations to SAP. ''^^ While the panel's 
occasional scrutiny of EPA's regulatory choices may technically exceed its sta- 
tutory mandate, the agency has learned to anticipate this tendency to stray afield. 
OPP now tries to frame specific questions for the panel to keep its focus on 
matters properly within its authority. ^^' Because SAP cannot be entirely blink- 
ered, however, EPA officials remain convinced that the panel's findings should 
not be binding on the agency. -^^^ 

E. Mechanisms for Centralized Risk Assessment 

The previous section examined the experience of the four agencies with 
panels of experts assembled to advise on the health risks of specific substances 
or activities. An accounting of this experience must acknowledge both successes 
and failures, whether measured by agency satisfaction or public respect for 
individual agency decisions or for institutions. The performance of EPA's Sci- 
ence Advisory Panel and FDA's continued support for expert advisory panels 
in several of its programs are convincing evidence that independent scientific 
review of agency risk assessments can be successful. An effort was made to 
identify features of advisory panels likely to contribute to agency and public 
satisfaction but such judgments inevitably have a strong taste of speculation. 

This section discusses three proposals for more drastic changes in the in- 
stitutional arrangements for making regulatory decisions. None of the proposals 
treated has actually been tried, so any evaluation necessarily is hypothetical. 
Indeed, none has yet been fully articulated; detailed descriptions of their me- 
chanical operation and working relationship with the current regulatory apparatus 
are lacking. Nonetheless, The proposals warrant treatment here because they 
have generated interest as solutions to perceived serious flaws in current agency 

1 . Background and Rationale. 

The proposals in question have a common feature: each would reassign 
responsibility for evaluating chemical hazards from the individual regulatory 
agencies to a new institution, a central body that would serve all four agencies. 
They include a proposal by the American Industrial Health Council, an industry 
group; ''^'^ a 1978 recommendation by the Office of Science and Technology 

320. Id. A recent study EPA for failing to make greater use of SAP. General Accounting 
Office, Delays and Unresolved Issues Plague New Pesticide Protection Programs (1980). 

321. Interview with Philip Gray, supra, note 294. 

322. Id.; interview with Steven Jellinek, supra, note 297. 

323. American Industrial Health Council, AIHC Proposal for a Science Panel, March 26, 
1980 [hereafter AIHC Proposal]. The proposal is amplified in a series of concept papers. See AIHC, 
"Toward a National Cancer Policy: A Framework for Sound Science in Federal Decision Making," 
August 18, 1981; AIHC, "A Proposal to Ensure a Sound Science Base for Governmental Policies 
and Regulations Relative to Chronic Health Hazards," August 18, 1981; AIHC, "Proposal for a 
Science Panel," March 26, 1980; AIHC, "AIHC Recommended Framework for Identifying Car- 
cinogens and Regulating Them in Manufacturing Situations," October 11, 1979. 


Policy, part of the Executive Office of the President; ^^"^ and a bill twice intro- 
duced by Virginia Congressman William Wampler. '''^-*' While these proposals 
visualize quite different institutional arrangements, they share one objective: 
determining the health risks of chemicals would no longer be the sole province 
of regulatory agencies. 

At the outset it is useful to outline the rationales that underlie these current 
expressions of dissatisfaction with traditional regulatory decisionmaking. Six 
different objectives are discernible in the proposals. Several of them, it will be 
seen, also support the use of expert advisory panels. 

a. Scientific independence . Suspicion of the objectivity of agency scien- 
tists is a recurrent theme in the proposals. The implicit argument is two-stage: 
First, decisionmaking about toxic chemicals can be divided into scientific risk 
assessment and policy design and selection. The choice of regulatory options is 
governed by statutory criteria and often involves determining the acceptability 
of risk, and is thus an appropriate function for politically attuned regulators. 
Risk assessment, however, demands scientific objectivity and should be inde- 
pendent from the regulatory process. ^^^ Second, agencies cannot be expected to 
provide this objectivity; their policy objectives may lead them to shade assess- 
ments of risk to support regulatory choices dictated by other considerations. ^^^ 
The solution, accordingly, is to place the risk assessment function outside the 
regulatory agencies. 

b. Scientific expertise. A companion rationale is the argument that risk 
assessment of potential carcinogens requires the nation's best scientists — who 
are not likely to be working in regulatory agencies. ^^^ 

c. Peer review. Another argument for independent risk assessment in- 
vokes the scientific ethic that new findings are not to be accepted until they are 

324. Office of Science and Technology Policy, Executive Office of the President, Identifi- 
cation, Characterization and Control of Potential Human Carcinogens: A Framework for Federal 
Decision Making, February I, 1979 [hereafter OSTP Proposal]. 

325. National Science Council Act of 1980, H.R. 638, 97th Cong., 1st Sess., 127 Cong. 
Rec. 2, 67 (1981). Originally introduced as H.R. 6521, 96th Cong., 2d Sess., 126 Cong. Rec. 22, 
901-02 (1980) [hereafter, Wampler bill]. 

326. " AIHC advocates that in the development of carcinogen and other federal chronic health 
control policies scientific determinations should be made separate from regulatory considerations 
and that such determinations . . . should be made by the best scientists available. ..." AIHC 
Proposal p. 1 . 

327. In discussing his bill before Congress, Representative Wampler pointed to FDA's threat- 
ened ban of saccharin in 1977 and commented that "It is my view, and I believe the evidence is 
getting stronger and stronger everyday, that our current crop of regulators are just not using good 
science to determine the hazards of substances in the market and or workplace, and may be even 
bending science to fit regulatory actions. . . ." 96th Cong., 2d Sess., 126 Cong. Rec. 36, EI097- 
8 (1980). 

328. "Having a single truly expert panel make these judgments for all federal agencies should 
provide a significantly greater degree of expertise than the various agencies could themselves as- 
semble." AIHC Book TAB E, p. 7. 


subjected to peer review. -^^^ The small numbers of top-rank scientists working 
for regulatory agencies, coupled with the disincentives to research and publication 
in that environment, effectively preclude such review within the agencies them- 
selves. Assessments of risk should be subject to scrutiny by scientists outside 
the agencies before becoming the basis for regulatory action. 

d. Consistency. A fourth justification is offered for creation of a new 
institution to assess the risks of chemicals. The premise is that agencies are often 
concerned about the same chemicals and should, logically, begin from a common 
scientific base in fashioning regulatory decisions. If each agency applies its own 
criteria in evaluating chemicals and performs its own assessment of the risks 
they pose, there is a high risk of inconsistency. Accordingly, a single body 
should have responsibility for assessing chemical risks for all four agencies. -^^^ 

e. Uniformity. Paralleling the consistency rationale is the argument that 
centralizing risk assessment will foster the development of uniform criteria for 
evaluating chemical risks. Despite the efforts of the ILRG^^' and the Regulatory 
Council, -^^^ the agencies cannot be relied on to produce a uniform set of standards. 

f. Chemical-by-chemical evaluation. Though not articulated, this rationale 
is implicit in the AIHC proposal. The argument, in substance, is that generic 
rules for evaluating chemicals discourage accurate risk assessment. Chemicals 
should be assessed individually, with attention to their mechanisms of action, 
metabolic fate, and other characteristics that conflict with necessarily general 
assumptions about carcinogenesis. ^^^ 

2. Content of the Proposals. 

It would not be productive here to describe in detail each of the three 
principal proposals for new risk assessment institutions. Familiarity with each 
proposal is not essential to evaluate their common premises or to identify potential 
problems of implementation. Furthermore, the authors of two of the proposals 
made clear that they were advocating a concept, not designing a set of procedures. 
Their descriptions leave many details to be filled in and many questions to be 
answered. Accordingly, this section does not present parallel summaries of the 

329. Representative Wampler commented that ''Until a scientific study has been reviewed 
by other scientists, its conclusions represent the opinions of the authors only. This has always been 
an unwritten rule in science. Why should a single unvalidated study applied to the regulatory arena 
be given a new and different connotation? If anything, the political consequences of the work make 
it even more susceptible to possible erroneous conclusions." Cong. Rec. Nov. 29, 1979, H-1 1394. 

330. Cf. the NTP panel which evaluated formaldehyde at the request of the IRLG agencies. 
See text accompanying notes 226-230. supra. 

331. Report of the Interagency Regulatory Liaison Group Work Group on Risk Assessment, 
"Scientific Bases for Identification of Potential Carcinogens and Estimation of Risks," 44 Fed. 
Reg. 39858 (1979). 

332. Regulatory Council, Regulation of Chemical Carcinogens. [1979] Chemical Regulation 
Reporter (BN A) 1081. 

333. See M. C. Throdahl, "Separating Science Definition from Legal Decision," Remarks 
to the Assembly of Life Sciences, National Academy of Sciences, October 12, 1981. p. 5; AIHC 
Proposal at 1 . 


three proposals. It briefly describes their source and basic contours, and then 
enumerates, in more general fashion, the many questions that would have to be 
resolved if a new institution were to be established to assess risks for all four 
regulatory agencies. 

a. OSTP proposal for bifurcated decisionmaking. In a 1978 paper several 
OSTP officials recommended the creation of a central risk assessment body to 
assure consistent identification, characterization, and assessment of potential 
human carcinogens. ^^"^ The report first distinguished two stages of regulatory 
decisionmaking: Stage I consists of identification of potential human carcinogens 
and quantification of the risk associated with exposure to those substances. Stage 
II consists of evaluation of regulatory options and their consequences. The OSTP 
report recommended that the Stage I process of scientific investigation and 
evaluation should be distinguished in time from the process of choosing regu- 
latory options and should be performed for all agencies by one body. 

The theme of the OSTP report was the need to separate the scientific and 
policy components of regulatory decisionmaking. It suggested that such a sep- 
aration will facilitate well-organized scientific inquiry, objective review of data, 
and uniform characterization of the risk associated with potential human carcin- 
ogens. ^^^ Furthermore, the report argued, adoption of a unitary decisionmaking 
framework would enhance public confidence in federal efforts to regulate po- 
tentially hazardous substances while also promoting efficiency and improving 
the quality of decisions. ^^^ 

The OSTP report recommended that Stage I (health effects assessment and 
risk analysis) should be placed under the auspices of an expanded National 
Toxicology Program, which should also develop uniform federal guidelines for 
epidemiological studies, animal tests, and quantitative risk assessment. ^^^ The 
report also suggested that all agencies conducting animal tests or using data 
generated from such tests should adopt uniform testing procedures. ^-^^ It further 
recommended that the nature of human exposure — sources, prevalence, dura- 
tion — should be determined as part of Stage I. Assessments of risk should take 
into account individual susceptibility and be quantified to the extent possible. ^^^ 

The OSTP report was silent on how Stage I would be set in motion, i.e., 
how a chemical would become a candidate for evaluation. Nor did it state 

334. OSTP ProposaL 

335. OSTP Proposal at 1. 

336. Id. 

337. Id. at 22. 

338. Id. at 10. The OSTP Proposal recommended that findings of carcinogenicity should be 
based on evaluation of all available evidence, in substance a "weight of evidence"" approach. Id. 
at 1 3. See also, McGarity. The Occupational Safety and Health Administration' s Generic Carcinogen 
Policy: Rulemaking Under Scientific and Legal Uncertainty 28, (unpublished manuscript 1981); 
Testimony of Elizabeth Anderson, Director, Office of Health and Environmental Assessment, EPA, 
Before the House Subcommittee on Operations, Research and Foreign Agriculture, June 23, 1981, 
p. 4. 

339. OSTP Proposal at 16-19. 


explicitly what weight was to be given to Stage I's conclusions concerning 
carcinogenicity and risk. The report's implication was that regulators who had 
been consciously excluded from the risk assessment process, would not be free 
to reexamine its results. ^"^^ Whether private parties interested in the outcome of 
regulatory decisions could challenge these results was left unclear. 

b. AIHC and Wampler proposals for a National Science Council. Unlike 
the OSTP report, which is more in the nature of a concept paper, two more 
recent proposals to establish new risk assessment institutions have been described 
in operational terms. Because these proposals use similar terminology and have 
several common features, they are summarized together here. 

On February 13, 1980 Congressman William Wampler introduced legis- 
lation to establish a National Science Council. ^"^^ Mr. Wampler's bill calls for 
the creation of a scientific panel, entirely independent of the regulatory agencies, 
to provide authoritative decisions on disputed questions of scientific fact.^"*^ The 
American Industrial Health Council had previously developed a similar proposal 
to create a central science panel to evaluate chemical hazards for all regulatory 
agencies. ^"^^ 

Both proposals stress the importance of uniform, consistent resolution of 
the scientific questions underlying regulatory decisions. Both emphasize the need 
to separate the scientific function of risk assessment from the policy making 
function of designing regulatory responses. Both rest on the assumption that the 
resolution of scientific issues in the context of individual regulatory proceedings 
creates the possibility of inconsistency. In statements supporting his bill Con- 
gressman Wampler has cited episodes, such as the abortive FDA-USDA plan to 
regulate nitrite, ^"^"^ to underscore the need for improved scientific analysis as a 
foundation for regulatory decisions. 

The Wampler bill and the AIHC proposal both represent attempts to remove 
ultimate responsibility for scientific judgments from the regulatory agencies. 
However, comparison exposes important differences between the two proposals. 
Wampler's bill provides that any party could request referral of scientific issues 
to the National Science Council. ^"^^ The AIHC proposal originally contemplated 
that the agency alone would have the authority to refer scientific questions to a 
central science panel. ^"^^ By its terms, the Wampler bill would apply only in 

340. /J. at 21. 

341. National Science Council Act of 1980, H.R. 638, 97th Cong., 1st Sess., 127 Cong. 
Rec. 2, 67 (1981). 

342. Wampler bill § 3(1 )(B)(2). 

343. American Industrial Health Council. "A Proposal to Achieve a Cohesive, National 
Cancer Policy," May 24, 1978. 

344. Congressional Record, Sept. 12. 1979, H7786-7791. 

345. See Wampler bill S 5 and § 19, which amends a variety of regulatory statutes to provide 
that only party to an adjudication conducted on the record may refer a question of scientific fact to 
the Panel. 

346. See AIHC. "A Proposal to Achieve a Cohesive National Cancer Policy," May 24, 
1978, p. 3. "Operating Procedures." 



formal adjudications, a limitation that would exclude most decisions concerning 
carcinogens, ^"^^ Under the AIHC proposal, an agency contemplating rulemaking, 
on its own initiative or upon request, could refer scientific questions for inde- 
pendent review by the panel. Because hazardous substances are regulated pri- 
marily through rulemaking, the AIHC proposal would have a far greater impact. 

Congressman Wampler's bill would make decisions of the National Science 
Council binding on any agency. ^"^^ An agency would not, however, be bound 
by the AIHC's science panel's conclusions. ^"^^ Rather, there would be a pre- 
sumption that the agencies would accept the Panel's findings in any subsequent 
regulatory action. 

The two bodies would also differ in composition and procedures. Wampler's 
bill contemplates that individual chemicals would be assessed by "advisory 
panels" to the National Science Council, which itself would be a standing body 
of fixed membership. ^^^ Membership on the advisory panels would, however, 
be restricted to members of the Council and no panel could have fewer than five 
voting members. ^^' The AIHC would allow its science panel to establish sub- 
committees or "working groups" which could include, or be composed largely 
of, outside experts. ^^^ Wampler's National Science Council would be composed 
of 15 full-time voting members serving two-year terms. ^^^ The AIHC recom- 
mends 15 part-time members who would serve for terms of three years. ^^"^ These 
divergent approaches to the question of full- versus part-time appointment appear 
to reflect different expectations about which status would be more likely to attract 
the best scientists. The AIHC approach assumes that scientists would be unwilling 
to devote their full time for a fixed term but might be persuaded to participate 
if appointment entailed a less burdensome commitment. 

Under the Wampler bill, the National Science Council would decide ques- 
tions of scientific fact after conducting "a hearing on the record. "^^^ Notice of 
the hearing would be given to all parties to the agency adjudication. The Council's 
proceedings would be formal, reinforcing its quasi-judicial character. -^^^ the AIHC 

347. Wampler bill § 5. "The Council shall decide . . . any question of scientific fact arising 
in any agency adjudication on the record. ..." See also Testimony of Peter Barton Hutt before 
the House Subcommittee on Department Operation, Research, and Foreign Agriculture, June 23, 
1981 p. 2. 

348. Wampler bill § 3(1 )(B). 

349. AIHC Proposal at 3. See also Memorandum of AIHC Counsel: "As noted above, the 
bill would require an agency to accept the Panel's finding and some might think it preferable to the 
AIHC Proposal in terms of expediting the regulatory process. On the other hand, the AIHC Proposal 
might meet with less opposition from the regulatory agencies because they would have an option 
of disagreeing with the Panel." 

350. Wampler bill § 7(a)(1). 

351. Id. 

352. AIHC Proposal at 6. 

353. Wampler bill §§ 15(a)(1)(A), 15(c)(1). 

354. AIHC Proposal at 4. 

355. Wampler bill § 5. 

356. See generally Wampler bill § 8. 


proposal, by contrast, makes no reference to "on the record" proceedings. An 
agency would simply announce referral of a scientific issue to the Panel^^'' and 
invite public comments. The less formal procedures visualized by the AIHC are 
consistent with the basic character of its proposal, i.e., to elicit high-level but 
non-binding judgments on scientific questions underlying regulatory decisions. 

A final important distinction is apparent in the time limits placed on the 
National Science Council by the Wampler bill. Under the bill, the Council would 
be expected to make a final report to the referring agency within 90 days of 
receiving a dispute. -^^^ The AIHC proposal, in contrast, expressly imposes no 
specific time limits on the Panel's decisionmaking,''^'^ reflecting a more realistic 
understanding of scientific and logistical complexities likely to confront any new 
advisory body. 

c. Responses to the proposals. The four agencies have responded critically 
to these proposals to separate and centralize the function of assessing the risk 
of chemicals that are candidates for regulation. •'^^ Other observers have also 
suggested flaws in the proposals. No attempt is made here to attribute specific 
criticisms, except where they are based on the distinctive experience of particular 
agencies. The purpose is to convey the basic criticisms of the Wampler bill and 
AIHC proposal. 

Perhaps the central criticism is that the theoretical distinction between risk 
assessment and the choice of regulatory options is oversimplified. ^^^ Like the 
proposals themselves, this argument is imprecise. Assessment of the toxicological 
data for a chemical and characterization of its propensity to induce cancer are 
surely different functions than determination of the costs of controlling exposure. 
Different disciplines are required to perform them. It is surely possible, in practice 
as well as in theory, to distinguish between estimating the risks posed by a 
chemical and deciding whether the risk estimated is acceptable. 

357. AIHC Proposal at 6. The AIHC proposes that upon referral of an issue to the Panel by 
a regulatory agency, the agency would publish a notice of the referral in the Federal Register, inviting 
the public and other agencies to submit relevant data to the Panel. 

358. Wampler bill § 10(a)(1). 

359. AIHC Proposal at 3. 

360. See Testimony of Elizabeth Anderson, supra, note 338: "EPA does not think that H.R. 
638 is an appropriate way to ameliorate the very real problem which it addresses, namely that the 
very best available scientific data and principles be used in setting health based standards and 
regulations." See also Testimony of Mark Novitch. Deputy Commissioner, U.S. Food and Drug 
Administration: "[T]his bill would . . . add another bureaucratic layer to the regulatory decision- 
making process with attendant waste of time and money." 

361. Testimony of Mark Novitch, supra, note 360: "[D]ecisions involving safety issues are 
only partly scientific: the scientific elements of a decision are closely related to its policymaking, 
legal, and regulatory elements. The answer to the question of whether a substance is harmful depends 
as much on the criteria or the standard one uses to judge harmfulness as on the technical facts. 
Assigning to a governmental body that is separate from the responsible regulatory agency the task 
of making a binding scientific decision in isolation from these other central elements could result 
in worse decisions, greater inconsistency, delayed administrative procedures, increased public con- 
fusion, and diminished public confidence." 



In two important respects, however, those who argue against institutional 
separation of the functions of risk assessment and regulatory policymaking make 
a valid point. First, they stress the large uncertainties that surround the central 
issues involved in regulating a chemical. -^^^ The agency responsible for deciding 
what exposure to permit or what costs to impose, the argument runs, must make 
what is ultimately a political judgment based on imperfect knowledge about 
every element. ^^^ For its decision to be politically acceptable, the agency must 
have responsibility for all of its components and the ability, throughout the 
process, to explore their relationship. -^^'^ 

This argument does not rebut all of the justifications for centralized risk 
assessment or preclude adoption of some features of the AIHC/OSTP proposals. 
It is not, for example, inconsistent with requirng an agency to consult with an 
expert advisory panel about the risks posed by a chemical before deciding whether 
or how to control it. The argument does, however, conflict with the concept of 
complete institutional separation of the risk assessment function. 

The second argument against bifurcation speaks to the internal process by 
which agencies reach decisions on complex matters. The contention is that this 
process is unavoidably an iterative one.^^^ Different specialists are called on 
repeatedly for analysis and advice as the agency identifies and considers new 
control options in attempting to reach a decision. While this description may 
overstate the fluidity of internal agency deliberations, it captures something of 
their unpredictable character. 

Regulators often must decide how to control a chemical to which humans 
are exposed in several different ways. This has implications for the analyses 
they must perform, as an example will reveal. Assume, first, that an agency 
desires quantitative estimates of the cancer risk for all exposed groups and, 
second, is permitted to balance control costs. As the agency explores the options 
for controlling exposure it often will discover new ones that it had not originally 
considered, or new combinations of controls, e.g., in the case of a pesticide, 
new restrictions on crop use combined with a new method of measuring residues 
and new instructions to applicators. If the agency has not anticipated this set of 
options, the independent scientific panel would not (unless it were clairvoyant) 
have estimated the risk remaining if they were imposed. It is unrealistic to expect 
that any agency, or the parties before it, will have thought of all of the options 
at the time the panel performs its original risk assessment. 

362. See, e.g.. Testimony of Mark Novitch, supra, note 360, pp. 6-7. 

363 . ' 'Testimony on Decision Making Processes for Food Safety and Quality Issues, ' ' Sherwin 
Gardner, Vice President, Grocery Manufacturers of America, Inc., June 23, 1981, p. 5. 

364. Testimony of Elizabeth Anderson, supra, note 338, p. 6: "[T]he risk assessment process 
is not accomplished in a single encounter as the Council might be forced to do, but rather it requires 
an ongoing interchange between scientists inside and outside the regulatory agencies who need to 
interpret the implications of the scientific evidence for regulatory decisions. Having the scientific 
evaluation process removed from the agencies would seriously impede those interactions and thus 
hinder the regulatory decision making process." 

365. See id. at 6. 


It is important to recognize that the assessment of risk and the choice of 
controls meet in the evaluation of exposures to a chemical. Quantitative assess- 
ment of the risks posed by exposure to a chemical requires information about 
the character and magnitude of human exposures. ^^^ But these are often the most 
uncertain elements of an agency's decision as well as the most likely to vary 
with the control measures selected. "^^^ The problem identified here could be 
avoided if the independent panel were only asked to characterize the chemical 
and estimate its effects under varying conditions of exposure. The agency would 
then be left to determine what exposure conditions prevailed, or could be achieved 
through regulation, and to apply the panel's dose-response model in performing 
its own quantitative estimates of risk. 

Other criticisms of the Wampler and AIHC proposals stress the logistical 
difficulties of meshing independent risk assessment with the internal workings 
of different agencies. ^^^ Experience suggests that it will be difficuh for any risk 
assessment body to meet rigid time limits. It is thus probable that agency decisions 
will be delayed by a requirement to consult with, or refer issues to, such a 
body.^^^ Given the current glacial pace of decisionmaking, however, the in- 
creased delay can hardly be viewed as a decisive argument against the proposals. 

EPA's experience with a comparable body — its own internal Carcinogen 
Assessment Group (CAG) — suggests another potentially more serious problem. 
CAG was created in 1976 to centralize agency evaluation of potential carcinogens 
for all of EPA's regulatory programs. ^^° It consists of approximately a dozen 
EPA scientists headed, until recently, by a distinguished academic researcher. 
As more EPA program units sought its assistance in evaluating chemicals, CAG 
became overburdened and its output grew unpredictable.^^' CAG failed to de- 
velop a mechanism for setting priorities and requests for its advice have some- 
times languished for many months. One consequence of this delay has been that 
• EPA's Office of Pesticide Programs began relying on its own scientists or on 
outside contractors, forfeiting whatever expertise CAG could provide. ^^^ 

Critics of the Wampler-AIHC proposals also challenge the assumption that 
the regulatory agencies have applied inconsistent criteria and reached inconsistent 
conclusions in evaluating different chemicals. ^^^ Until EPA recently declined to 
initiate regulation of formaldehyde^^'* after the CPSC had proposed to ban UP 

366. Id. at 4. 

367. See, e.g., Merrill, Consumer Products at 1349, 1366. 

368. Testimony of Elizabeth Anderson, supra note 338, p. 7; Testimony of Peter Barton Hutt, 
Esq., pp. 6-9. 

369. Testimony of Mark Novitch, supra note 360, p. 10. 

370. Risk Assessment in the Federal Government at 105. 

371. Interview with Steven Jellinek, supra note 297. 

372. Id. 

373. Testimony of David D. Doniger, Natural Resources Defense Council, Before the House 
Subcommittee on Departmental Operations Foreign Agriculture, June 23, 1981, p. 2. See also 
Testimony of Elizabeth Anderson, supra note 321, p. 9; Testimony of Mark Novitch, supra note 
345, p. 10. 

374. See [1982] Chemical Regulation Reporter 1207 (BNA). 


foam insulation, ^^^ it would be difficult to cite an example of inconsistency. To 
be sure, the agencies have often selected different control options or specified 
different exposure limits for the same chemicals, but such disparities have typ- 
ically reflected differences in exposure (and thus in estimated risk), or, more 
commonly, in statutory requirements. The claim that a single risk assessment 
body is needed to avoid inconsistencies among the IRLG agencies is difficult to 

A more convincing argument for centralization is that it might expedite, 
and in any case would reduce the administrative costs of, decisionmaking when 
two or more agencies contemplated regulation of the same substance. If it made 
sense to seek the advice of independent scientists on the risks it posed, there is 
a good deal to be said for having one panel perform this function for all interested 
agencies. This was the rationale for the IRLG's request to the National Toxi- 
cology Program to empanel a group of scientists to evaluate the CUT study of 
formaldehyde. ^^^ Although the panel's confirmation of formaldehyde's carcin- 
ogenicity does not appear to have stilled industry criticism nor produced agree- 
ment among the agencies, the concept merits continued investigation. 

The Wampler bill has been criticized on several other grounds. The pre- 
scription that members of the National Science Council be full-time appointees 
has been termed unreliable. "^^^ The bill's conception of a quasi-judicial body 
making binding determinations of scientific issues seems excessively formal. 
The bill conjures up images of presentations orchestrated by attorneys, punctuated 
by legal challenges, and based on rules of evidence. ^^^ Its limitation to agency 
adjudications would effectively exclude most controversial regulatory proceed- 
ings. Telling criticims of the Wampler and AIHC proposals are, thus, numerous. 
The former proposal, in particular, seems ill-adapted to the institutional and legal 
context in which regulatory agencies now operate. The operational flaws apparent 
in these proposals, however, should not entirely obscure the fundamentally sound 
objectives for regulatory decisionmaking that they embody. The proposals cor- 
rectly emphasize the analytical distinction between the assessment of risk and 
the choice of regulatory options. They recognize the value of independent sci- 
entific review of agency assessments of carcinogenicity and risk. And they stress 
the need for government-wide consistency in the scientific bases for health safety 

F. Generic Criteria for Regulating Carcinogens 
1 . Introduction 

As Professor Scalia has observed, ^^^ the 1970's saw a veritable explosion 
of rulemaking by federal agencies. Administrative law scholars and practitioners 

375. See Merrill, Consumer Products at 1357-61. 

376. See text accompanying notes 236-40 supra. 

311. See, e.g., Testimony of Peter Barton Hutt, Esq., before the House Subcommittee on 
Departmental Operations, Research, and Foreign Agriculture, June 23, 1981, p. 9. 

378. Id. at 5-6. 

379. A. Scalia, Vermont Yankee: The APA, The D.C. Circuit, and the Supreme Court, 1978 
Sup. Ct. Rev. 345, 376-77. 


had previously advanced several reasons to support such an approach to regu- 
latory policymaking. ''Generic rulemaking ""^^^ promised efficiencies in resolv- 
ing individual cases that would otherwise relitigate recurrent issues. ^^' Informal 
rulemaking was also touted as a less expensive, more accessible process for 
diverse interests to make known their views. -^^^ Judges endorsed rulemaking as 
a means of clarifying agency policy and injecting predictability into agency 
decisions. ^^"^ 

Two of the agencies studied for this report — OSHA and FDA — have at- 
tempted to use rulemaking to establish generic standards for carcinogens. ^^"^ 
Neither agency's experiment has yet been implemented. OSHA promulgated 
comprehensive criteria for evaluating and regulating workplace carcinogens in 
1980, but litigation delayed implementation and the Reagan administration has 
indicated its intention to rethink these regulations.'^^'' FDA's initiative, originally 
proposed in 1973, still remains unfinished. ^^^ The two agencies' approaches, 
though resting on common scientific and legal promises, are not parallel. FDA's 
proposal addresses a narrow though complex set of issues encountered in reg- 
ulating a single class of products, animal drugs; by contrast, OSHA's generic 
carcinogen regulations span all of the issues it has encountered in setting health 
standards that pose a risk of human cancer. The two efforts illustrate both the 
advantages and difficulties of attempting to resolve complex scientific and policy 
issues by general rule. 

To evaluate the success of these efforts — and the utility of generic rule- 
making in health and environmental regulation generally — one should under- 
stand why they were undertaken. Conversations with individuals who conceived 
the regulations discussed in this section provide some insights, but motivation 
is difficult to confirm after the fact and, in large organizations, even more difficult 
to attribute. Lacking complete information about original objectives, I assume 
that FDA's and OSHA's rulemaking initiatives had one or more of the following 

(a) Rulemaking would provide a means of resolving general issues of agency 
policy, which could not readily be addressed in proceedings concerning specific 

380. This common term is a non sequitur, for all rulemaking is by its nature generic and 
attempting to generalize from particular cases or derive principles that govern more than a single 

381. See, e.g.. C. Ames & S. McCracken. Framing Regulatory Standards to Avoid Formal 
Adjudication: The FDA as a Case Study, 64 Cal. L. Rev. 14. 60-61 (1976). 

382. T. McGarity. The Occupational Safety & Health Administration' s Generic Carcinogen 
Policy: Rulemaking Under Scientific and Legal Uncertainty 13 (unpublished paper 1981). See also 
40 Fed. Reg. 22950 (1975). revoked 40 Fed. Reg. 40520 (1975). and superseded 40 Fed. Reg. 
40682 (1975), illustrating FDA's regard for informal rulemaking as a means of making policy. 

383. See, e.g., H. Friendly. The Federal Administrative Agencies: The Need for Better Def- 
initions of Standards, 75 Harv. L. Rev. 863, 878-79 (1962). 

384. The CPSC made an abortive effort to adopt criteria for evaluating potential carcinogens 
by general rule. See Merrill, Consumer Products at 1297-1303. 

385. See 45 Fed. Reg. 5001 (1980): 47 Fed. Reg. 187 (1982). 

386. See Merrill and Hutt at 496-506. 


substances. A companion objective may have been to elicit broader participation 
in agency policymaking. 

(b) Rulemaking would narrow the issues in proceedings concerning specific 
substances. I would include within this rationale the objective of avoiding hear- 
ings altogether, i.e., using rulemaking to support "administrative summary judg- 

(c) Rulemaking would permit agency management to clarify, and force 
internal adherence to, regulatory policy. 

(d) Rulemaking would demonstrate the agency's concern about a problem 
that seemed unresponsive to case-by-case attack. 

This list of purposes for generic rulemaking is both artifical and incomplete. 
It fails to capture the nuances of judgment that motivate collective decisions to 
undertake new initiatives as well as the political pressures under which agencies 
operate — or believe they operate. But without at least some hypothesized ob- 
jectives for generic rulemaking it is difficult to evaluate the success of any effort. 

Of the four objectives suggested here, only the second is clearly subject to 
legal constraints. Because there is a substantial body of case law dealing with 
administrative summary judgment and because this objective was central to 
OSHA's attempt to establish general rules for regulating carcinogens, some 
attention to these constraints is appropriate before discussing the two agency 
initiatives in detail. 

2. Administrative Summary Judgment 

OSHA's generic carcinogen regulations represent an attempt to apply es- 
tablished administrative law principles in a new regulatory context. The agency 
relied on three Supreme Court decisions, Storer Broadcasting, ^^^ FPC v. Tex- 
aco,^^^ and Weinberger v. Hynson, Westcott & Dunning,^^^ each of which sus- 
tained the denial of an evidentiary hearing to an applicant or licensee who 
challenged the agency's refusal to authorize commercial activity. Together, these 
cases confirm that an agency may avoid (or limit) evidentiary hearings by pro- 
mulgating general criteria for decision whose applicability in individual cases is 
evident from undisputed facts. ^^ 

FDA's successful use of administrative summary judgment in the Hynson 
case involved the interpretation of scientific data and therefore is the most ger- 
mane. Under the 1962 Drug Amendments to the Federal Food, Drug, and Cos- 
metic Act,^^' any new drug must be shown by "adequate and well-controlled" 
clinical studies to be effective for its labeled uses.^^^ To avoid repeated litigation 
over the effectiveness of hundreds of pre- 1962 drugs, FDA adopted regulations 

387. United States v. Storer Broadcasting Co., 351 U.S. 193 (1955). 

388. FPC V. Texaco, 377 U.S. 33 (1964). 

389. Weinberger v. Hynson, Westcott, & Dunning, 412 U.S. 609 (1972). 

390. J. Mashaw and R. Merrill, Introduction to the American Public Law System 483 (1975) 
[hereafter Mashaw and Merrill]. 

391. 21 U.S. C. § 321-92. 

392. 21 U.S.C. § 355(d). 


prescribing the features that cHnical tests must have to meet the statutory standard. 
The agency specified that to be entitled to a hearing, a manufacturer must submit 
studies conforming to the regulations.'''^-^ FDA then proceeded, summarily, to 
withdraw marketing approval for a large number of pre- 1962 drugs whose man- 
ufacture failed to marshall evidence of the specified kind.'*'^'^ 
The Supreme Court upheld FDA's general procedure: 

If FDA were required to hold automatically a hearing for each product 
whose efficiency was questioned by the NAS-NRC study, even though 
many hearings would be an exercise in futility, we have no doubt that it 
could not fulfill its statutory mandate to remove from the market all those 
drugs which do not meet the effectiveness requirements of the Act.'*'^-'' 

The Court did not mention the agency's inclusion of a provision for waiver of 
the rule's requirements; it did, however, emphasize that the agency's ability to 
deny a hearing depends on the obviousness of the deficiencies in an applicant's 
studies. ^^^ 

The cases suggest some limitations on the use of generic rulemaking to 
narrow individual disputes. '''^^ The effective scope of administrative summary 
judgment, however, is more likely to be a function of context than of legal 
principle. An agency apparently may use rulemaking to resolve recurrent issues 
of fact or of policy that arise in proceedings involving similar activities. The 
rules upheld range from those expressing policy judgments, ^'^'^ to rules disposing 
of complex factual disputes, such as the design of test protocols for new med- 
icines. ^^^ The need to provide an opportunity for waiver of general rules may 

393. The regulations purported to provide flexibility by permitting the FDA Commissioner 
to accept a nonconforming study if the drug's sponsor convincingly demonstrated that the specific 
criteria for "adequate and well controlled" investigations should not apply. 21 C.F.R. § 
314.1 1 l(2)(5)(ii)(a). For a discussion of the judicial and regulatory foundation for FDA's use of 
"administrative summary judgment," see Mashaw and Merrill at 459-89. 

394. Ames and McCracken, supra, note 381. at 17. 

395. Weinberger v. Hynson. Westcott & Dunning. 412 U.S. at 621. 

396. Id. In order to prevail at a hearing, an applicant must furnish evidence from "adequate 
and well-controlled investigations" since the Court "cannot impute to Congress the design of 
requiring, nor does due process demand, a hearing when it appears conclusively from the applicant's 
pleadings that the application cannot succeed." Justice Douglas stressed, however, that this applies 
only to regulations which are precise. While, under some criteria, a mere reading of the study 
submitted will indicate whether it is deficient, other criteria will call for the exercise of discretion 
or subjective judgment in determining the adequacy of the study. Qualitative criteria, such as 
"adequate" or "suitable," .96'^, f.g., 21 C.F.R. § 130. 12(a)(5)(ii)fajf2j(/j, "do not lend themselves 
to clear-cut definition and it may not be possible to tell from the face of a study whether the standards 
have been met. Thus, it might not be proper to deny a hearing on the ground that the study did not 
comply with this regulation." 412 U.S. at 621, n.I7. 

397. See Mashaw and Merrill at 488-89; McGarity, supra note 382 at 29. See also Phar- 
maceutical Manufacturers Ass'n v. Finch, 307 F. Supp. 858 (D. Del. 1970). 

398. E.g., United States v. Storer Broadcasting Co., supra, note 376. 

399. E.g., Weinberger v. Hynson, Westcott & Dunning, supra, note 378. See also The Use 
Of Generic Rulemaking to Resolve Environmental Issues in Nuclear Power Plant Licensing, 61 Va. 
L. Rev. 869 (1975). Id. 


depend on subject matter; a rule that purports to resolve technical issues in a 
rapidly changing field, for example, may have to contemplate more departures 
from the norm.^"^ 

3. Experiments in Generic Rulemaking 

a. FDA's "sensitivity of method" proposal. FDA's latest attempt to es- 
tablish generic rules for regulating carcinogens is expressed in a 1979 proposal 
outlining criteria for evaluating animal drugs/^' To understand this initiative, 
one must be familiar with the statutory context. 

The 1958 Food Additives Amendment to the FD&C Act was interpreted 
as prohibiting FDA approval, for use in animals produced for human food, of 
any drug that had been shown to cause cancer; this was a "food additive" use 
barred by the Delaney Clause. '^°- In 1962 Congress amended the clause, as it 
applied to animal drugs, to permit FDA to approve the use of a carcinogen if 
the agency were convinced that "no residue" of the substance would be "found" 
in edible tissues of treated animals. "^^-^ The amendment specified that FDA was 
to prescribe the analytical method, or assay, for verifying the absence of residues. 

Progressive improvements in analytical chemistry have since confirmed that 
no drug administered to animals is ever eliminated entirely. FDA has thus come 
to recognize that the key issue under this provision is the detection, not the 
occurrence, of drug residues. Accordingly, FDA has proposed criteria for ap- 
proving methods to search for drug residues.'^^'^ Rather than gear those criteria 
to the changing capability of chemical analysis or to some fixed numerical limit, 
the agency defined its standard in terms of risk. It proposed that any assay 
approved for controlling a carcinogenic drug must be capable of measuring 
residues that present more than "insignificant" risk of cancer. And the agency 
specified a 1 in 1,000,000 lifetime risk of cancer as a quantitative measure of 

400. See, Starr v. FAA, 589 F.2d 307, 312 (7th Cir. 1978). "No statute or regulation limits 
the specific subject matter areas in which petitions for exemptions may be entertained. [However] 
it is reasonable to assume that the administrator can identify certain matters to be more appropriately 
subject to exemption procedures than others." 

401. 44 Fed. Reg. 17070 (1979). See Merrill and Hutt at 496-506. 

402. Diethylstilbesterol (DES) posed a dilemma for the FDA. While the Delaney Clause 
prohibited approval of new DES uses and products, the drug's approval could not be withdrawn 
since it was a "prior sanctioned" substance exempt from retroactive application of the Food Additives 
Amendment. Merrill and Hutt at 484. 

403. The so-called "DES proviso" provides: 

[Ejxcept that this proviso shall not apply with respect to the use of a substance as an 
ingredient of feed for animals which are raised for food production, if the Secretary finds 
(i) that, under the conditions of use and feeding specified in proposed labeling and rea- 
sonably certain to be followed in practice, such additive will not adversely affect the 
animals for which such feed is intended, and (ii) that no residue of the additive will be 
found (by methods of examination prescribed or approved by the Secretary by regulations, 
which regulations shall not be subject to subsections (f) and (g) in any edible portion of 
such animal after slaughter or in any food yielded by or derived from the living animal. 

21 U.S.C. § 360a(c)(3)(D). 

404. 44 Fed. Reg. 17070 (1979). 


"insignificance." If a drug sponsor could provide a detection method capable 
of measuring residues posing any greater risk,"*^^ FDA in effect would ignore 
residues it could not find. 

This description illustrates the thrust of FDA's proposal, but it fails to reveal 
importance of FDA's experiment in policymaking or the complexity of the issues 
it confronted. The agency had been wrestling with the problem of prescribing 
analytic methods for animal drugs since 1962. Its proposal is an attempt to 
establish durable criteria that take into account differences in the risk posed by 
different drugs. "^^^ 

FDA's proposal is intended to serve the same functional purpose as the 
FCC's no-more-than-five stations rule, i.e., to resolve generically one of several 
issues that arise repeatedly in license application proceedings. But the two pol- 
icies differ in important ways. Though the FCC's policy embodied several con- 
siderations respecting concentration of media ownership, it captured them in a 
simple rule whose application would in most cases be obvious. "^^ By contrast, 
to determine what level of residues poses no more than 1 in 1 ,000,000 risk of 
cancer, a sponsor of an animal drug must perform a sequence of costly exper- 
iments, which are prescribed by FDA's proposal. For example, the FDA proposal 
specifies criteria for evaluating the potential carcinogenicity of animal drugs, 
standards for chronic feeding studies, and comparative metabolism studies to 
determine whether humans are likely to be exposed to different residues than 
those produced by the animals in which the drug caused cancer. "^^^ Each of these 
proposed requirements resolves complex issues about which scientists continue 
to disagree. Many resist the notion that these issues can or should be resolved 

• FDA's proposal addresses other issues that are no less complex but may 
be more appropriate for generic resolution. First, the proposal frankly embraces 
quantitative risk assessment both as a technique for evaluating carcinogens and 
as a criterion for determining permissible exposure levels. As I have observed 
in Part I of this report, the issues underlying these choices, though difficult, are 
not likely to be sharpened by scientific advances in the near term, and sound 
administrative policy dictates that they be resolved consistently — at least within 
individual agency programs. FDA's proposal also selects one mathematical model — 
the linear model — for extrapolating from high-dose data from animal experiments 

405. In earlier proposed regulations. 38 Fed. Reg. 19226 (1973). FDA prescribed a mathe- 
matical extrapolation model and designated an acceptable level of risk. It specified that residues 
posing no greater than one in 100 million lifetime individual risk of cancer could be ignored. After 
an "avalanche" of comments, however, the agency accepted a risk level of one in a million. 42 
Fed. Reg. 10412 (1977). The 1977 regulations were struck down on procedural grounds and later 
revoked by FDA. 43 Fed. Reg. 22675 (1978). 

406. Merrill and Hutt at 496-97. 

407. To be sure, there might be cases in which it was unclear whether an applicant's interest 
in another station constituted "ownership" for purposes of the rule, but these were likely to be rare. 

408. 44 Fed. Reg. 17096ff (1979); 44 Fed. Rep. 17084ff (1979); and 44 Fed. Reg. 17083ff 


to low-dose human exposures. The choice of an extrapolation model is, similarly, 
appropriate for generic resolution. 

b. OSHA's generic carcinogen policy. While a large number of chemicals 
have been identified as potential occupational carcinogens, OSHA has established 
standards for fewer than twenty. Each proceeding to set a standard for a toxic 
substance has been protracted, ranging from fifteen months to over five years. ^^^ 
According to some observers, one cause of OSHA's slow pace was the enormous 
records that its rulemakings generated, which in turn reflected the controversy 
of its proposals and the scope of the issues on which it entertained comment. "^'^ 
By 1976, agency officials had become convinced that if the standard-setting 
process were not expedited, decades would elapse before it could regulate all 
currently suspected carcinogens. 

On October 4, 1977, therefore, OSHA proposed general criteria and pro- 
cedures for setting standards for potential occupational carcinogens."^^ ^ Its an- 
nounced objective was to obviate relitigation of the same issues in successive 
standard- setting proceedings. It accordingly proposed to resolve, as matters of 
scientific fact or administrative policy, the principles by which it would identify, 
classify, and regulate chemicals that posed a cancer risk to humans. The agency 
thus hoped to forestall reexamination of these principles in subsequent rulemak- 
ings."^'^ Under the proposal, once OSHA officially classified a chemical as a 
workplace carcinogen, it would immediately issue an Emergency Temporary 
Standard and, within 60 days, propose adoption of a Model Proposed Standard 
requiring reduction of employee exposure to the lowest feasible level. "^'^ The 
central issues in the ensuing rulemaking hearing would be classification of the 
substance as a workplace carcinogen; determination of the lowest feasible level 
of exposure; and identification of less hazardous substitutes. "^'"^ 

409. The following table illustrates the duration of OSHA's process: 

Time Elapsed from Proposal 

Chemical Regulated 

to Final Standard 


6 mos. 

14 Carcinogens 

6 mos. 

Vinyl Chloride 

5 mos. 


3 years, 4 mos. 

Coke Oven Emissions 

3 years 

Cotton Dust 

1 Vi years 


9 mos. 


5 mos. 


9 mos. 

See McGarity, supra note 382, at 


410. Id. 

411. 42 Fed. Reg. 54148 (1977). 

412. 42 Fed. Reg, 54154 (1 


OSHA . . . has determined that it is necessary at this time to the fulfillment of its statutory 
objectives, to reshape the size and content of its rulemaking proceedings. . . . 

413. 42 Fed. Reg. 54148 (1977). 

414. 42 Fed. Reg. 54171 (1977). 


Development of OSHA's final carcinogen policy proved as difficult as any 
of its rulemakings for individual chemicals. The informal hearing lasted from 
May 16, 1978 until June 25, 1978 and generated a transcript of over 8,500 
pages. "^'^ The full rulemaking record eventually exceeded 250,000 pages. OSHA 
itself offered more than forty-five witnesses, and a larger number of witnesses 
appeared on behalf of other participants. The final policy occupied more than 
280 Federal Register pages of preamble and a dozen of regulatory text."^'^ The 
regulations are an attempt to "codify" many of the scientific and policy premises 
that underlie OSHA's earlier standards and thus to discourage continued debate 
over what the agency regards as accepted principles of carcinogen regulation. 
In this respect, the final regulations adhered to the conception of the proposal, 
but they incorporated several changes intended to limit their dispute-foreclosing 

OSHA's generic regulations discuss in detail the criteria it will use in 
evaluating different types of evidence concerning a chemical's carcinogenicity. 
For example, the regulations state that the agency will accept positive results in 
well-designed epidemiological studies as establishing "the qualitative inference 
of carcinogenic hazards to workers.""*'^ By contrast, negative human evi- 
dence,"*'^ "will be considered by the Secretary" only if: 

(i) The epidemiological study involved at least 20 years' exposure 
of a group of subjects to the substance and at least 30 years' observation 
of the subjects after initial exposure; [and] . . . 

(iii) The group of exposed subjects was large enough for an increase 
in cancer incidence of 50% above that in unexposed controls to have been 
detected at any of the predicted sites. . . ."^'^ 

OSHA's regulations similarly attempt to foreclose issues often encountered 
in evaluating annual experiments: 

(g) ... Positive results for carcinogenicity obtained in mammals 
exposed to high doses of a substance will be used to establish the qualitative 
inference of carcinogenic hazards to workers. . . . 

(h) ... No determination will be made that a "threshold" or "no- 
effect" level of exposure can be established for a human population exposed 
to carcinogens in general, or to any specific substance. 

(i) ... Results based on the induction of benign or malignant tumors, 
or both, will be used to establish a qualitative inference of carcinogenic 
hazard to workers. . . .^■^^ 

The agency's approach to evidence purporting to rebut the foregoing propositions 
parallels its treatment of non-positive human studies. '^^' 

415. 45 Fed. Reg. 5001, 5008 (1980). 





§ 1990.143(a) 


29 C.F.R. 

§ 1990.143(c) 


29 C.F.R. 

§ 1990.144. 


29 C.F.R. 

§ 1990.143(a) 


29 C.F.R. 

§ 1990.143. 


OSHA's preamble reiterates the agency's long-standing position that the 
Act requires it to reduce employee exposure to substances posing a cancer risk 
to the lowest level compatible with industry survival and achievable by fore- 
seeable technology /^^ The agency expressly declines to engage in cost-benefit 
analysis to establish such exposure limits /^"^ Furthermore, its policy embodies 
a decisive preference for engineering controls and work practices to achieve 

This policy has been applied by OSHA in every health standard for 
potential carcinogens. Yet, time and again similar arguments have been 
made on the relative effectiveness, cost, the need for employer discretion, 
and other issues. OSHA has carefully reviewed all relevant aspects of the 
record on this issue and concludes that substantial evidence supports the 
continued preference for engineering and work practice controls. Accord- 
ingly, OSHA has foreclosed further consideration of this issue except under 
the amendment provisions. . . ^^"^ 

4. Impact of FDA and OSHA Rulemakings 

a. FDA proposal. It is probably premature to evaluate FDA's "sensitivity 
of method" proposal as an exercise in generic rulemaking. Like OSHA, FDA 
is under new management whose commitment to the agency's 1979 proposal is 
not known. Even in its present posture, however, the proposal permits some 
speculation about the advantages and limitations of generic rulemaking in the 
regulation of carcinogens. 

FDA's proposal addresses several complex technical issues as well as the 
policy question of what criteria the agency should apply in approving residue 
assay methods, but it pales in scope beside OSHA's 300-page document. Thus 
the potential for mistaken resolution of issues, i.e., for enunciating general rules 
that do not fit numerous cases, is probably less. Nor was FDA's original proposal 

422. 29C.F.R. § 1990.144. 

423. While asserting that it has routinely considered the ability of an industry to absorb the 
costs of more stringent standards, OSHA declined, on three grounds, to follow a cost-benefit approach 
in setting exposure limits: 

First, the state-of-the-art of quantitative risk assessment techniques particularly for 
carcinogens, that are crucial to cost-benefit analysis, is too crude and unreliable to permit 
the refined estimates necessary to determine risk reduction at various exposure levels. . . . 

Secondly, as a matter of policy, efficiency criteria alone are not appropriate because 
they ignore equity considerations. The economic savings from less protective regulation 
accrue to industry in the form of higher profits and consumers in the form of lower prices. 
But the costs are borne by workers through increased industrial illness and death rates. 
OSHA believes that it would be inappropriate for an agency statutorily required "to assure 
so far as possible every working man and woman in the Nation safe and healthful working 
conditions" to ignore such equity considerations. 

Finally, OSHA believes that the correct interpretation of the Act requires the agency 
to reduce worker exposure to carcinogens to the lowest feasible level since no safe level 
can be shown. 

45 Fed. Reg. 5001, 5239 (1980). 

424. 45 Fed. Reg. 5001, 5222 (1980). 


inspired by a desire to lay the basis for summary judgment in disputes over 
specific drugs. Agency lawyers later became aware that final "sensitivity of 
method" regulations might support summary disposition of some proceedings, 
but the driving force behind the initiative was the judgment of FDA managers 
that the agency needed a new answer to the question of how hard to search for 
drug residues /^^ In other words, the purpose was to change rather than codify 
agency policy. 

It remains unclear whether the proposal has unified thinking about carcin- 
ogenic animal drugs within FDA, often a desired by-product of rulemaking. 
There is evidence that, in evaluating new compounds, reviewers in FDA's Bureau 
of Veterinary Medicine and Bureau of Foods had begun to rely on the principles 
underlying the proposal even before it was published. On the other hand, the 
time required to generate agreement on a final document may suggest that internal 
disputes are not readily resolved by public rulemaking. "^^^ 

The background of FDA's 1979 proposal dramatizes both the duration of 
its gestation and the costs of attempting to resolve the issues it raises by general 
rule. The effort began in the early 1970's, when the discovery of DES residues 
in beef liver, using unapproved analytical methods, convinced FDA officials 
that the agency needed a new approach to regulating carcinogenic animal drugs. '^^^ 
In 1973 the agency published a proposed regulation, similar in broad outline to 
the 1979 version, on which it invited public comment. "^^^ Four years later, after 
protracted internal review of the comments and debate about the propriety of 
the approach proposed, FDA published a final "sensitivity of method" regu- 
lation.'*^^ Drug manufacturers successfully challenged this regulation in court on 
the ground that its contents differed too substantially from the 1973 proposal, 
thus sending the agency back to the drawing board. "^^^ 

The 1979 proposal was the product of two more years' work preparing 
responses to earlier comments, to the court's mandate, and to advances in un- 
derstanding of drug metabolism and quantitative risk assessment. For the past 
year there have been periodic reports in the trade press that a final regulation 
would soon be published, but as of this writing, the policy FDA first advanced 
in 1973 remains only a proposal. 

425. The prospect of summary disposition of individual cases loomed less important because 
FDA has not been heavily burdened by administrative hearings on the approvability of animal drugs. 
In its evaluation of applications for the approval of new products, for example, the agency has 
confronted only one demand for an evidentiary hearing since the passage of the 1962 amendments. 
But requests for hearings have been common when FDA has sought to revoke approval of already 
marketed drugs, and agency lawyers foresaw that it would confront many more hearing requests in 
the future. 

426. See note 405 supra. 

All. See Hess & Clark, Div. of Rhodia, Inc. v. FDA, 495 F.2d 975 (D.C. Cir. 1974). 

428. 38 Fed. Reg. 19226 (1973). 

429. 42 Fed. Reg. 10412 (1977). 

430. Animal Health Institute v. FDA, [1977-1978 Transfer Binder] Food Drug Cos. L. Rep. 
(CCH) § 38,154 (DDC 1978). 


This account obscures one important product of FDA's effort. Through the 
three Federal Register documents FDA has illuminated the complex issues posed 
by Congress' instruction to be sure that "no residue" of a carcinogenic drug 
"is found" in human food. Furthermore, it has elicited broad, if not unanimous, 
agreement that a standard calibrated to the carcinogenic potency of individual 
drugs is a sensible solution to the problem.'*^' 

b. OSHA's generic carcinogen policy. Implementation of OSHA's reg- 
ulations has been temporarily derailed. After initially proposing changes in the 
policy in response to the Supreme Court's benzene decision,"^^^ OSHA withdrew 
them and announced that more extensive revisions would soon be offered."^^-^ 
Later the agency proposed rulemaking, announcing its intention to reconsider 
the policy and staying implementation of the priority-setting requirements pre- 
viously described.'*^'* Accordingly, it is too soon to assess the regulations' impact 
on OSHA's program for setting health standards. A preliminary assessment is 
nonetheless in order. 

Participants in OSHA's rulemaking saw its objectives in different terms. 
Some may have hoped that the process would provide a forum for ventillating 
the scientific issues involved in regulating carcinogens and result in an accom- 
modation of competing views. "^^^ Others believed that establishment of a "cancer 
policy" would help OSHA respond to sharp criticism of its slow pace in estab- 
lishing health standards. "^^^ Finally, establishment of generic criteria, coupled 
with a priority-setting mechanism and prescribed rulemaking schedule, was ex- 
pected to expedite the process for setting standards for individual chemicals. 

OSHA's policy differs in scope and in practical effect from the regulations 
upheld in prior cases as supporting summary judgment. The regulations address 
the full range of scientific and policy issues encountered in previous rulemakings; 
they describe the process by which OSHA will evaluate chemicals before com- 
mencing rulemaking; and they speak to remedial issues, such as the content of 
standards and the desired means of compliance with them. While OSHA's reg- 
ulations are thus more ambitious than similar generic rules, they are likely to 
have a less draconian impact. 

The FCC, the FPC, and FDA relied on generic standards to avoid adjudi- 
catory hearings on individual license applications. "^^^ OSHA, by contrast, is 

431. See 42 Fed. Reg. 10412 (1977). 

432 . Industrial Union Dept . , AFL-CIO v . American Petroleum Institute , 448 U . S . 607 ( 1 980) . 
OSHA proposed changes in the regulations at 46 Fed. Reg. 4889 (1981). 

433. The carcinogen policy's effective data was first delayed until the end of March, 1981. 
Subsequently, the amendments prompted by the API decision were withdrawn, 46 Fed. Reg. 1900 

434. 47 Fed. Reg. 187 (1982). 

435. See McGarity, note 368 supra. 

436. Interview with Grover Wrenn, former OSHA Director for Health Standards (June 26, 

437. The FCC, the FPC, the FDA all relied on generic regulations to denying claims to 
adjudicatory hearings ostensibly guaranteed by statute and in some cases by Constitution as well. 
See the Communications Act § 309, 47 U.S.C. § 309; FPC: The Natural Gas Act § 717c(e), 15 
U.S.C. § 717c(e); Food, Drug, and Cosmetic Act § 505(e), 21 U.S.C. § 355(e). 


attempting to narrow the scope of future rulemaking proceedings. This distinction 
should make its approach less vulnerable to attack on constitutional grounds. 
Moreover, OSHA's regulations purport only to narrow the issues it will consider 
in subsequent rulemakings rather than to foreclose hearings altogether. "^^^ 

. . . [P]roceedings for individual substances under this Part shall be 
limited to consideration of the following issues: 

(a) Whether the substance, group of substances or combination of 
substances subject to the proposed rulemaking is appropriately considered 
in a single proceeding; 

(b) Whether the substance . . . meets the definition of a potential 
occupational carcinogen . . . including whether the scientific studies are 
reliable . . . 

(i) The determination of the lowest feasible level to control exposures 
. . . primarily through the use of engineering and work practice controls 
including technology and economic consideration; . . . 

(k) Whether suitable substitutes are available for one or more uses 
of [the substance] . . . ; and 

(1) Whether the provisions of . . . [the model standard] are appro- 
priate. . . ."^^^ 

In partial response to comments that its proposal would foreclose evidence on 
issues that are still subject to scientific debate, OS HA also included two types 
of "waiver" provisions. '^'^^ 

Any assessment of OSHA's carcinogen policy must start from the assump- 
tion that the regulations are valid. "^"^^ Even making this assumption, it is doubtful 
whether OSHA's regulations will materially improve its efficiency in setting 
standards for workplace carcinogens. OSHA cannot escape rulemaking, includ- 
ing informal hearings, to set a standard for any chemical; its regulations expressly 
leave numerous issues open for individual standard-setting proceedings. '^'^^ More 
important, the regulations are studiously ambiguous about the fate of arguments 
or evidence that challenge the agency's general conclusions but do not meet its 
criteria for "consideration." Notably, they do not state that such submissions 
will be excluded from the rulemaking record or ignored by the Assistant Sec- 

438. OSHA Draft at 108-09. 

439. 29 C.F.R. § 1990.143. 

440. See 29 C.F.R. §§1990.144; 1990.145(b). The latter regulation would permit any par- 
ticipant to submit "substantial new evidence" on issues that are otherwise presumptively foreclosed. 
"For the purposes of this Part, 'substantial new evidence' is evidence directly relevant to any provision 
of this Part and is based upon data, views, or arguments which differ significantly from those 
presented in establishing this Part, including amendments thereto." 

441 . For a discussion of OSHA's assertion that it possesses general rulemaking authority, see 
OSHA Draft at 81-85. See also Air Line Pilots Association, Int'l v. Quesada, 276 F.2d 892, (D.C. 
Cir. 1960), which upheld generic rulemaking where the same issue would be repeatedly presented 
in individual adjudicatory or licensure proceedings. See generally, Judicial Review of Generic 
Rulemaking: The Experience of the Nuclear Regulatory Commission, 65 Geo. L.J. 1295 (1977); 
Note, The Use of Generic Rulemaking to Resolve Environmental Issues in Nuclear Power Plant 
Licensing, 61 Va. L. Rev. 869 (1965). 

442. See text accompanying notes 432-33, supra. 


retary. This omission is understandable. Rigid notions of admissibility are in- 
compatible with informal rulemaking, and one expects that OSHA would in fact 
accept many written submissions that arguably do not meet its requirements. 

While OSHA employees do not expect the regulations will result in the 
exclusion of evidence, they predict that the Assistant Secretary will be able to 
dispose of much evidence perfunctorily, thereby avoiding the delays of previous 
rulemakings. '^'^^ I remain skeptical that the resulting savings will be significant. 
First, there will be evidence and arguments whose eligibility for full "consid- 
eration" will be a matter of dispute. For example, a test conducted to demonstrate 
that a substance produces metabolites in animals that are not produced in man 
will ordinarily require careful evaluation before it can be rejected. "^"^ The obli- 
gation to respond to pertinent comments"^"^^ will, I believe, lead OSHA to devote 
time to submissions that it might technically be entitled to ignore. A standard- 
setting proceeding represents too big an investment for the agency to take the 
risk of neglecting comments that an reviewing court might later regard as ma- 

It is equally doubtful whether OSHA's regulations will have a major impact 
on the conduct of rulemaking hearings. Under the agency's practice, the presiding 
officer is not responsible for preparing an initial or recommended decision; his 
job is to keep order and be sure the record is complete. Specifically, he is not 
expected to rule on the admissibility of arguments or evidence. ^^ Again, the 
risk of erroneously excluding testimony is likely to induce acceptance of every- 
thing "for what it is worth." It should be noted here, as Professor McGarity 
has reported, that the duration of hearings was never a major cause of delay in 
OSHA's rulemaking process. 

Even if hearings were eliminated entirely, the pace at which OSHA proposes 
and finalizes regulations would probably not accelerate appreciably. . . . 
[T]he time consumed in actual hearings is a miniscule proportion of the 
time consumed between the moment the agency is asked to promulgate a 
standard and the time it issues a final standard. The glacial pace at which 
the agency has issued health standards has deeper reasons, including White 
House pressure to delay issuance of standards, inadequate and/or incom- 
petent staff in the Standards Division, poor management, rapid turnover in 

443. Wrenn Interview, supra, note 436. 

444. 29 C.F.R. § 1990.144(c) provides: 

Arguments that differences in metabolic profiles can be used to demonstrate that a chemical 
found positive in an experimental study in a mammalian species would pose no potential 
carcinogenic risk to exposed workers will be considered by the Secretary only if the evidence 
presented for the specific substance subject to the rulemaking meets specified criteria. 

The criteria then listed are a series of strict documentation and interpretation requirements indicating 
the agency's general skepticism for the proposition discussed. 

445. E.g. United States v. Nova Scotia Food Products Corp., 568 F.2d 240 (2d Cir. 1977); 
Portland Cement Ass'n v. Ruckelshaus, 486 F.2d 375 (D.C. Cir 1973), cert, denied, 417 U.S. 921 
(1974); Mobil Oil Corp. v. FPC, 483 F.2d 1238 (D.C. Cir 1973). 

446. See 29 C.F.R. § 1911. 15-. 16. 


leadership, poor interaction between managers in OSHA and scientists in 
the National Institute for Occupational Safety and Health, poor interaction 
between OSHA and its lawyers in the Office of the Solicitor of Labor, and 
the requirement that OSHA perform lengthy regulatory analyses before it 
issues standards. While streamlining the hearings will probably enhance the 
agency's efficiency somewhat, it is not likely to aid the agency greatly in 
reaching its goal of increasing its output of health standards, if indeed, that 
is any longer a goal for the agency. "^^ 

Finally, the efficiencies provided by OSHA's regulations could be small 
even if the agency were never again confronted with evidence or arguments on 
any of the issues they presumptively foreclose. Despite the agency's protestations 
that it has repeatedly heard the same arguments, "^^ examination of its prior 
rulemakings does not reveal a preoccupation with old issues. There have been 
intense disputes above the interpretation of specific studies — disputes that are 
certain to continue — and there has routinely been debate over the cost and 
technical feasibility of control measures — matters expressly left open by the 
regulations for case-by-case resolution. 

Despite this generally skeptical assessment, one can identify ways in which 
the regulations might aid OSHA's decisionmaking. First, the regulations should 
help focus the issues in standard-setting proceedings by indicating what types 
of arguments OSHA will find convincing. "^^ While individuals cannot be pre- 
vented from introducing evidence that the regulations purport to preclude, par- 
ticipants may not spend resources assembling information that they expect OSHA 
to dismiss. Second, the regulations describe the process by which OSHA expects 
to identify and set priorities for regulating the potentially large number of work- 
place carcinogens. The preparation of the carcinogen regulations might have 
been expected to yield other benefits, as well. The internal debate which is 
necessary to complete a major rulemaking can sometimes improve agency man- 
agement. The need to produce a final product may precipitate agreement on 
disputed issues, or misunderstandings of agency policy that previously escaped 
attention. Furthermore, the consideration of public comments can be genuinely 
educational for agency officials. It can only be a matter of speculation, however, 
whether OSHA's effort to promulgate its generic carcinogen regulations has had 
or will have any of these effects. 

G. Interagency Coordination 

1 . Introduction . 

This section explores issues raised by proposals to coordinate federal reg- 
ulation of carcinogens. The desirability of interagency coordination has been 
discussed for several years, and efforts to bring about consistency and cooperation 

447. McGarity, supra, note 382 at 26. 

448. See 45 Fed. Reg. 5001, 5108-15 (1980). 

449. Given the fact that those affected by agency rules have finite resources for challenging 
policies, they are likely to focus most of their energies on those issues for which the agency remains 
open to persuasion. 



among the four agencies were begun as early as 1976. In 1977 the heads of the 
four agencies formed the Interagency Regulatory Liaison Group (IRLG)."^^^ FDA 
Commissioner Kennedy and EPA Administrator Costle saw advantages in co- 
ordination for both agencies and manufacturers and users of chemicals. These 
included uniformity of criteria for testing and evaluation, predictability of de- 
cisions, efficiencies in inspection and enforcement, and enhanced government 
sensitivity to the economic costs of disparate control measures. Another less 
often articulated rationale for the formation of the IRLG was the belief that the 
agencies had to respond to claims that their decisions were inconsistent and their 
choice of targets largely adventitious. 

The IRLG became the primary vehicle for coordination among the four 
(later five) agencies. "^^^ It functioned through eight working groups of agency 
scientists and regulators assigned to such tasks as criteria for evaluation of 
scientific evidence, inspections, enforcement, and test standards development. 
Later, the Carter Administration also formed the Regulatory Council, which 
assumed as one of its first responsibilities the development of a government- 
wide "cancer policy. '"^^^ The activities of the IRLG working groups are too 
diverse to describe or evaluate here;"^^^ many had little impact within the agencies, 
much less outside. 

The IRLG coordination umbrella was folded during the early months of the 
Reagan Administration, which substituted a new, but much less elaborate, co- 
ordinating apparatus, headed by the Office of Science and Technology Policy 
in the Executive Office of the President'*^'^ and consisting of the heads of the 
five IRLG agencies. Precisely what role this new group will play in overseeing 
government evaluation and regulation of carcinogens cannot now be predicted. 
This section, therefore, will focus on activities of the IRLG and on specific 
proposals for coordinated regulation of specific chemicals that drew attention 
during its period of operation. 

2 . The Concept of Coordinated Regulation 

The formation of the IRLG increased interest in the possibility of joint or 
coordinated proceedings involving chemicals subject to regulation by two or 
more agencies. One of the original IRLG work groups was assigned responsibility 
for identifying chemicals of potential interest to two or more agencies. 

The notion of coordinated regulatory proceedings has intuitive appeal; it 
appears to promise efficiencies for both government regulators and private parties 

450. Regulation of Toxic and Hazardous Substances: Interagency Agreement, 42 Fed. Reg. 
54856 (1977); see also 43 Fed. Reg. 7174 (1979). 

45 1 . The Food Safety and Quality Service of the Department of Agriculture joined the IRLG 
in 1979. See 44 Fed. Reg. 39858 (1979). 

452. See Statement on Regulation of Chemical Carcinogens; Policy and Request for Public 
Comments, 44 Fed. Reg. 60037 (1979). 

453. The IRLG formed eight different working groups responsible for formulating policies 
for coordination in such areas as monitoring of research laboratories, establishment of test guidelines, 
and development of regulatory actions. 

454. See 1 Inside O.M.B., No. 3, p. 1, February 1, 1982. 


and to assure that different agencies will not arrive at inconsistent decisions. 
While agencies have often cooperated in regulating specific chemicals,'*^'' how- 
ever, the operational meaning of coordination has remained undefined. The 
following discussion identifies three goals of coordinated regulation and then 
examines, by hypothesis and example, ways in which the four agencies might 

a. Coordinated rulemaking. "Coordinated rulemaking'' has been used to 
epitomize what is meant by coordinated regulation, but this description is not 
ideal. "Coordination" may include a variety of cooperative activities that do 
not lead to, or require, public rulemaking. Furthermore, the term "rulemaking" 
is underinclusive because some statutes do not permit regulation of chemicals 
through promulgation of rules. For example, an action by EPA to cancel a 
pesticide's registration must offer the full range of procedural safeguards asso- 
ciated with formal adjudication under the APA.'^''^ In other contexts, an agency 
may have a choice between rulemaking and adjudication. But most attempts to 
regulate carcinogens have taken the form of rulemaking. Three of the four 
programs studied here — the exception being EPA's regulation of pesticides — 
utilize rulemaking. The reader should, however, remain cognizant of the pos- 
sibility that some chemicals cannot be regulated through rulemaking, collabo- 
rative or otherwise. This possibility is both a reminder, and a product, of the 
variety of statutory procedures that govern the four agencies. 

b. Rationales for coordination. The idea of interagency coordination in 
the regulation of carcinogens is appealing for several reasons. Most chemicals 
regulated as carcinogens have fallen within the jurisdiction of two or more 
agencies. Indeed, few chemicals produced in this country would not fall within 
the jurisdiction of at least two agencies: OSHA, which would be responsible for 
protecting workers engaged in production of a chemical, and the agency (or 
agencies) responsible for its commercial uses. As a practical matter, some po- 
tential carcinogens may be of concern to only one agency. A few chemicals are 
no longer in production, and thus American workers are therefore no longer 
exposed to them, but they remain in the environment and thus, accessible to 
humans. "^^^ An example is aflatoxin, a carcinogenic mold that occurs naturally 
on many grains, which is of interest only to FDA. Such exceptions are not 
unimportant, but virtually all chemicals that currently are candidates for regu- 
lation as carcinogens fall within the jurisdiction of two or more agencies. Ac- 
cordingly, it seems logical to explore ways of coordinating agency efforts to 
evaluate and regulate carcinogens. 

455. Perhaps the most notable example was the collaborative efforts of FDA, EPA, and the 
CPSC in regulating aerosol uses of chloroflurocarbon propellants, whose emission is believed to 
damage the ozone layer and increase the risk of skin cancer from ultraviolet radiation. See 42 Fed. 
Reg. 24535 (1977). 

456. See FIFRA § 6(d), 7 U.S.C. § 136d (1978). 

457. For example, FDA still monitors the food supply for the occurrence of residues of 
pesticides whose registrations have long since been cancelled by EPA. The tolerances FDA enforces 
are generally those established by EPA. 


Three primary objectives of coordination can be identified: assurance of 
consistent decisions; administrative economy, with savings both for agencies 
and for members of the pubHc; and coherent implementation of controls. While 
these are laudable goals, they require clarification. 

(i) Consistent decisions. A few years ago a Wall Street Journal editorial 
chastised the three agencies for reaching inconsistent decisions on the same 
substance. "^^^ The Journal pointed out that OSHA, FDA, and EPA had established 
different exposure limits for acrylonitrile, a plastic that has many useful appli- 
cations but also the capacity to cause cancer in laboratory animals. OSHA had 
set an occupational health standard of 2 ppm; EPA had established a higher limit 
on acrylonitrile as an air pollutant; and FDA had attempted to ban its use in the 
manufacture of plastic bottles because minute amounts could leach into the 
beverages they contained. For the Wall Street Journal, these disparate standards 
demonstrated that federal regulation of toxic chemicals was in disarray. 

To be fair, the editorial writer did not attribute the problem to any failure 
of the agencies to cooperate, but rather to Congress' failure to develop a coherent 
"cancer policy." The example does, however, reveal the importance of refining 
consistency as a rationale for coordinated regulation. A study of their actions 
on acrylonitrile reveals that the three agencies were in agreement on the scientific 
issues common to their decisions. Each concluded that the experimental data 
demonstrated that acrylonitrile is a carcinogen. The agencies also generally 
agreed in their assessment of the risk to humans who inhaled or ingested acry- 
lonitrile, i.e., no level of exposure could be assumed to be without risk. To be 
sure, each agency looked to different data in evaluating the exposures for which 
it was responsible. Most exposure issues, i.e., whether it is likely to occur and 
at what levels, are parochial rather than common. Since any quantitative as- 
sessment of risk turns on predicted levels of exposure, estimated risks are also 
likely to differ. However, the choice of an extrapolation model for quantifying 
the risk of different exposures is potentially a common issue. The choice of 
model is likely to be a function of regulatory philosophy, on which government- 
wide consistency might be expected, or of a model's compatibility with the data, 
which ordinarily are the same for all agencies. '^^^ 

The exposure limits that EPA, OSHA, and FDA set for acrylonitrile were 
determined only in part by their assessments of risk. The primary determinant 
of FDA's decision that the bottles could not be sold was the "no risk" standard 
of the Food Additives Amendment. "^^^ Similarly, OSHA's exposure limit was 
dictated by its interpretation of the "feasibility" criterion of the OSH Act"^^^ the 
issue least susceptible of ' 'consistent' ' resolution is what exposure to a carcinogen 
is acceptable. Exposures vary widely and governing laws understandably set 
different standards. It is accordingly understandable, not surprising, that agencies 
might choose different exposure limits. 

458. "A Low Growth Microcosm," Wall Street Journal, October 10, 1978. 

459. See Risk Assessment in the Federal Government at 34-40. 

460. See Monsanto Co. v. Kennedy, 613 F.2d 947 (D.C. Cir. 1979). 

461. See Occupational Exposure to Acrylonitrile, 43 Fed. Reg. 45762 (1978). 


While this discussion suggests the limits of the consistency rationale, it also 
reveals areas appropriate for coordination. It is one thing for government agencies 
to set different limits to different hazards. It is quite another for government to 
provide conflicting answers to such questions as whether a chemical is capable 
of causing human cancer. 

(ii) Process efficiencies. Because agencies engaged in regulating the same 
chemical must explore many of the same scientific questions, it would seem 
sensible to develop procedures that permit these questions to be resolved once. 
These questions are likely to surround interpretation of the data relating to the 
chemical's behavior in mammalian systems and its carcinogenicity. The inter- 
pretation of a single animal experiment may give rise to a dozen contested 
questions. "^^^ A unitary procedure for resolving such questions has considerable 
appeal. The agencies should save time, and each should gain the aid of the others 
in assembling, presenting, and evaluating data. Private parties, who might oth- 
erwise expect to participate in each proceeding at which the chemical's fate is 
at issue, would have to make their case only once. Scientists whose studies are 
in dispute and other experts who might testify would be spared repetitive ap- 

There may also be efficiencies in unified exploration of other issues. 1 have 
suggested earlier that exposure issues typically differ for each agency, but ex- 
posure data are scarce and costly to obtain. Regulators must therefore often rely 
on estimates and inferences based on other measured exposures. "^^^ A study of 
exposures in one setting may reveal valuable information about how the chemical 
behaves generally. Furthermore, even when exposure issues differ for each agency, 
they may not be decisive in determining what steps should be taken to control 
a chemical. If a chemical were of concern to only two agencies, for example, 
the toxicity and risk issues might dominate, making it efficient to resolve them 
in a common proceeding. Furthermore, exposure issues may occasionally be 
nearly identical for different agencies. An example is sodium nitrite, which is 
regulated by both FDA and USDA. 

(iii) Coordinated control measures. Coordination can help prevent conflict 
in the implementation of controls imposed by two or more agencies . For example , 
when EPA and FDA restricted the production and use of chloroflurocarbon 
propellants in consumer products, they agreed that an immediate ban would be 
economically disruptive and provide slight health benefits. Accordingly, it was 
decided that manufacturers of consumer products would be permitted to exhaust 
their stocks and to use up existing supplies of the propellant. While EPA ordered 
production terminated almost immediately, FDA delayed its ban on commercial 
distribution of chloroflurocarbon-containing drugs and cosmetics to correspond 
to the time required to exhaust existing supplies."^^^ Their cooperation, however, 

462. For a discussion of the variety of issues potentially raised by the results of animals 
studies, see IRLG, Scientific Bases for Identification of Potential Carcinogens and Estimation of 
Risks, 44 Fed. Reg. 39857, 39862-869 (1979). 

463. See Risk Assessment in the Federal Government at 20-22. 

464. See Fully Halogenated Chlorofluoroalkanes: Proposed Prohibitions and Notice of CPSC 
Action, 42 Fed. Reg. 24535 (1977). 


did not require joint rulemaking. Interagency meetings prior to the publication 
of separate proposals focused the timing issue, on which both agencies invited 
comments which were later jointly evaluated. "^^^ 

(iv) Other motives for coordination. There are other reasons why agencies 
interested in the same chemical may wish to collaborate. Through such efforts 
a small agency like the CPSC, which lacks the scientific capabilities of EPA 
and FDA, can gain help in evaluating complex data. This, surely, was one reason 
that the CPSC sought the help of the IRLG and National Toxicology Program 
in evaluating formaldehyde. "^^^ Another appeal of coordination is strength of 
numbers. The IRLG was in part inspired by the belief of the four agency heads 
that health and safety regulation was under critical attack. The alliance provided 
strength in dealing both with outside groups and with a White House staff that 
was becoming increasingly concerned about the cost of regulatory controls. 

3. Obstacles to Coordination. 

The procedural impediments to joint rulemaking can be illustrated by a 
hypothetical example involving a current candidate for regulation, formaldehyde. 
A panel of government scientists, assembled to review the bioassay performed 
by CUT, have concluded that formaldehyde is carcinogenic for rodents. Relying 
on this judgment, the CPSC in 1980 proposed to ban UF foam insulation, and 
early this year voted to make the ban final, although this action is based as well 
on formaldehyde's acute effects. "^^^ Formaldehyde has many commercial uses. 
It is a component of other consumer products and of some products under FDA's 
jurisdiction, workers engaged in manufacturing formaldehyde are exposed to its 
fumes. Finally many uses of formaldehyde are subject to TSCA. Accordingly, 
all four regulatory agencies might someday wish to limit formaldehyde's pro- 
duction, marketing, or use. If each agency acted independently, they would 
duplicate efforts, burdening both industry and government. Each agency's stat- 
ute, however, imposes different procedural requirements for regulatory action. 
If the four agencies desired to collaborate, it would be necessary to determine 
how these diverse procedures could be meshed to meet the requirements of each 

In this analysis I assume that the four agencies do not wish to effect a 
complete ban on formaldehyde, i.e., that some uses justify the attendant risk of 
cancer. "^^^ Let us assume that the tentatively agreed-upon objectives are a selective 
ban on certain uses of formaldehyde and specified limits on permitted exposures. 
A second simplifying assumption is that three of the four agencies are content 

465. Id. 

466. See text accompanying notes 238-243 supra. 

467. See 46 Fed. Reg. 1 1 188 (1981); [1982] Chemical Regulation Reporter (BNA), p. 1219. 

468. Such a trade-off would not be uncommon for some prescription drugs, notably anti- 
tumor agents, whose benefits are thought to outweigh the risk of cancer in some populations. OSHA's 
benzene standard accepted the possibility that some workers would continue to get cancer from 
exposure to benzene, by implication conceding that the costs of eliminating exposure altogether 
were excessive. 


to rely on a single statutory remedy. The exception is FDA, whose primary 
statute prescribes different procedures for regulating drugs and cosmetics, the 
two classes of products within its jurisdiction that contain formaldehyde. "^^^ 

A threshold issue is whether the four agencies should conduct a joint reg- 
ulatory proceeding or conduct parallel but separate proceedings. Neither the APA 
nor any of the four agencies' statutes provided for joint rulemaking proceedings, 
but I assume that such Congressional silence does not betoken disapproval so 
long as all of the requirements of the APA and of pertinent regulatory statutes 
are met by whatever procedure is followed. "^^^ This assumption has two corol- 
laries. First, any joint proceeding would have to satisfy the minimum require- 
ments of the most demanding agency statute. I refer to this as the "highest 
common denominator" problem. A consequence is that in any joint proceeding 
one or more agencies might have to observe procedural requirements that exceed 
the requirements of their own statutes. This would be a practical obstacle to 
collaboration unless the agencies saw advantages in joint proceedings that out- 
weighed the increased procedural cost. The second corollary is that each agency 
would have to make its own factual findings and draw its own legal conclusions 
to discharge the responsibilities delegated to it.'^'^' 

The "highest common denominator" problem could be avoided if the four 
agencies simply conducted contemporaneous proceedings conforming to the re- 
quirements of their respective statutes. But this approach might sacrifice some 
of the advantages of a joint proceeding. For example, it could force members 
of the public with interests in the decisions of more than one agency to prepare 
duplicate submissions, to recall witnesses, and to make identical arguments to 
several tribunals. Furthermore, an agreement to conduct parallel proceedings 
could dissolve if one agency became diverted by other pressures. An agency's 
incentive to adhere to a common timetable would be diminished if it realized 
that only its proceeding would be disrupted by a delay. A joint proceeding, on 

469. FDA is empowered to withdraw approval for any new drug that is found to be unsafe, 
see 21 U.S.C. § 355(e). or to demand proof of safety for any drug, previously marketed with an 
approved new drug application, whose safety comes into question. Before withdrawing or refusing 
approval, however, the agency must afford the manufacturer an opportunity for a formal evidentiary 
hearing. Id. The agency's authority over cosmetics is less comprehensive. It may initiate court 
enforcement proceedings against a cosmetic that it considers adulterated because it contains a "poi- 
sonous or deleterious substance which may render it injurious to health." See 21 U.S.C. § 361(a). 
On occasion it has also attempted to ban hazardous ingredients from cosmetics through rulemaking, 
a procedure whose legality is not clearly established. Under the agency's view of its authority, 
however, it could accomplish such a ban through rulemaking conducted in accordance with the 
informal requirements of 5 U.S.C. § 553. 

470. It is difficult to perceive the basis for legal objection to such a procedure if each statute's 
procedural requirements were punctiliously followed. 

471. Since the authority to regulate is conditioned on the existence of facts — or at least 
findings — specified in any agency's statute, and the authority is delegated to the official or body 
also authorized to make the requisite findings, an agency head presumably could not act on the basis 
of findings made by someone else — even a fellow regulator. Of course, the agency head could give 
weight to the findings of his colleagues in other agencies so long as the record before him supported 


the Other hand, would be continuously vulnerable to the changing priorities of 
each agency and hostage to the least efficient. 

With the exception of thd FD&C Act, each agency statute prescribes pro- 
cedures for rulemaking beyond the minimum required by the APA."^^^ The rule- 
making procedures outlined in the Toxic Substances Control Act (TSCA), for 
example, fall somewhere between the notice and comment requirements of sec- 
tion 553 and the more formal procedures outlined by sections 556 and 557 of 
the APA. TSCA requires the EPA to provide an opportunity for an informal 
hearing with verbatim transcription, with opportunity for cross-examination and 
rebuttal if necessary to resolve disputed issues of material fact."^^-^ 

The OSH Act mandates slightly different hybrid procedures for setting 
occupational health standards. Notably, the Act specifies that an informal hearing 
must be held if objections to OSHA's proposed rule are filed. Under OSH A 
practice, this hearing is presided over by an Administrative Law Judge, is tran- 
scribed verbatim, and often includes lengthy cross-examination of witnesses. "^^"^ 

As recently amended, the CPS Act prescribes the most complex procedures 
for rulemaking that would govern any of the four agencies. When it contemplates 
establishing a product safety standard or ban, the CPSC must first publish an 
advance notice of proposed rulemaking. "^^^^ Furthermore, the CPSC must pre- 
viously have obtained the views of a Chronic Hazards Advisory Panel. "^^^ Once 
a proposal is published, the CPS Act requires the Commission to "give interested 
persons an opportunity for the oral presentation of data . . . [and] a transcript 
shall be kept of any oral presentation. '"^^^ CPSC hearings are more informal 
and typically briefer than those conducted by OSHA. 

FDA would have a choice of means for restricting the use of formaldehyde 
in prescription drugs. Through informal rulemaking it could declare formalde- 
hyde an adulterant under section 501(a)(3) of the FD&C Act.'*^^ If the Com- 
missioner determined that a generic approach were not appropriate, however, 
he would have to proceed against individual products under section 505(e)'^^^ of 
the FD&C Act. That section empowers FDA to withdraw approval of a new 
drug application (a product license) if it determines that the drug, as composed, 
is unsafe. The procedure prescribed requires notice and an opportunity for a 
formal evidentiary hearing. By contrast, to curtail the use of formaldehyde in 
cosmetics, FDA could simply promulgate a rule declaring all products containing 

472. FIFRA prescribes formal adjudication, 7 U.S.C. § I36d (1978); the Occupational Safety 
and Health Act prescribes a varient of informal rulemaking that includes an opportunity for an oral 
hearing, 29 U.S.C. § 655(b) (1976); and the Consumer Product Safety Act mandates a similar 
version of hybrid rulemaking, 15 U.S.C. § 2058(a)(2) (1976 & Supp. Ill 1979). 

473. 15 U.S.C. § 2605(c)-(d) (1976). 

474. See OSHA at 27-30. 

475. See Merrill, Consumer Products at 1271 -IS, 1374-75. 

476. Id. at 1374-75. 

477. 15 U.S.C. § 2058(a)(2) (1976 & Supp. Ill 1979). 

478. 21 U.S.C. § 351(a)(3). 

479. 21 U.S.C. § 355(e) (1976). 


the chemical adulterated under section 601(a) of the FD&C Act.'^^^ Only the 
minimum procedures of section 553 of the APA would apply to such an action. 

If each agency opted for the least formal procedures available to it, the 
minimum procedures that would all satisfy statutory requirements would be 
informal rulemaking, coupled with a transcribed hearing that afforded some 
opportunity for cross-examination. It is not obvious that each agency would 
prefer the least formal procedures. Assume, however, that each of the four 
agencies would agree to regulate formaldehyde through the most expeditious 
procedures compatible with its own statute, joint action might proceed as follows: 

Once the agencies had agreed on the outlines of their ultimate policy, they 
could publish either a single advance notice of proposed rulemaking or simul- 
taneous ANPR's. Although some of the statutes prescribe information that must 
be included in any advance notice, '^^' it should not be difficult for the four 
agencies to develop a joint document, outlining the substance of the proposed 
rules and describing the issues they raise. To satisfy the minimum requirements 
of the most demanding statute, the CPS Act, at least sixty days would have to 
be allowed for public comments, but this period is shorter than the time likely 
to be allowed. "^^^ 

Because TSCA, the OSH Act, and the CPS Act all require an opportunity 
for some sort of informal oral hearing on proposed rules, any joint rulemaking 
proceeding would have to include a hearing. TSCA requires an opportunity for 
rebuttal and cross-examination on disputed issues of material fact."^^^^ Since TSCA 
would probably be relied on to ban production of formaldehyde for uses forbidden 
by other agencies, these requirements would have to be observed in resolving 
the issues raised by any proposal. With respect to issues raised by limits proposed 
under other laws, a legislative-type hearing that permitted oral presentation would 
presumably suffice. Whether it would make sense to conduct different types of 
hearings on the distinct issues seems doubtful. It probably would be simpler to 
permit any proposed ban, as well as exposure and use restrictions, to be contested 
at a single TSCA-sufficient hearing, even though this would open each factual 
issue to cross-examination. 

Furthermore, it should be noted, the issues to be explored at such a hearing 
would be as broad as required by the most elaborate statutory formula. The 
statutory standards applicable in our example range from the standard for de- 
termining whether a substance adulterates cosmetics, which FDA has construed 
as essentially precluding any risk to consumers. "^^"^ By contrast, TSCA imposes 

480. 21 U.S.C. § 361(a) (1976). 

481. See the 1981 amendments to the Consumer Product Safety Act, discussed in Merrill, 
Consumer Products at 1374-75. 

482. See id. Except in an emergency, the customary practice of all four agencies has been to 
afford at least sixty days, and often considerably longer, for public comments on proposals to ban 
or restrict important substances. 

483. See 15 U.S.C. § 2605(c) (1976). 

484. 21 U.S.C. § 361(a) (1976). See generally Merrill and Hutt at 626-34. 


an "unreasonable risk" standard, which arguably required EPA to weigh all 
measurable effects of curtailing production or use of a chemical. "^^^ The amended 
CPS Act appears now to call for some approach cost-benefit analysis /^^ In short, 
any joint hearing would have to address a wide range of issues, not all of which 
would be relevant to all agencies. 

Following such a hearing, the four agencies would develop final individual 
rules based on the common rulemaking record. Both OSHA and the CPSC would 
be required, within sixty days of the hearing, to issue a rule or make a deter- 
mination that no rule should be issued. "^^^ In practice, both agencies have often 
postponed this deadline. Neither FDA nor EPA under TSCA is subject to time 
restrictions on the promulgation of final rules. Some of the statutes prescribe 
the form and content of final regulations."*^^ If each agency were promulgating 
its own rule, however, any difference among these requirements should not pose 
a problem. 

Additional problems could be raised by collaboration in evaluating the 
rulemaking record and in drafting final rules. First, although not likely to succeed, 
it could be argued that post-comment period conversations between staff members 
from one agency, e.g., EPA, and decisionmakers in another agency, e.g., the 
CPSC, would constitute inappropriate ex parte communications. While the in- 
hibitions first suggested in Home Box Ojfice^^^ have been qualified in subsequent 
decisions, "^^^ this issue cannot be ignored. If staff members of one had acquired 
a reputation as critics of a chemical, manufacturers of cosmetics might be ex- 
pected to challenge any arrangement that afforded them preferential access to 
decisionmakers at FDA. Yet to preclude off-the-record communications among 
the four agencies after the comment period would frustrate collaboration. 

A second problem raised by post-hearing cooperation involves the respon- 
sibility of decisionmakers for their decisions. Courts rarely seriously entertain 
accusations that an official formally responsible for a decision did not make it."^^^ 
But most such challenges have failed in the context of accusations that the 
decisionmaker had abdicated to his subordinates. In the present context, the 
Administrator of EPA, for example, might rely on the work of employees of 
other agencies. While one would not expect challenges asserting improper sub- 
delegation of responsibility to succeed, they are likely to be made. 

485. See generally C. Davies, S. Gusman, and F. Irwin, Determining Unreasonable Risk 
Under the Toxic Substances Control Act, An Issue Report of the Conservation Foundation (1979). 

486. See Merrill, Consumer Products at 1374. 

487. See 15 U.S.C. § 2058(a)(2) (1976 & Supp. Ill 1979), and 29 U.S.C. § 655(c) (1976). 

488. For example, the Consumer Product Safety Act requires the CPSC to make and set forth 
a series of findings respecting the nature and degreee of risk and alternative regulatory approaches 
before banning a product or substance. See Merrill, Consumer Products at 1280. 

489. Home Box Office, Inc. v. Federal Communications Comm'n, 567 F.2d 9 (D.C. Cir. 
1977), cert, denied, 434 U.S. 829 (1978). 

490. See, e.g.. Action for Children's Television v. Federal Communications Comm'n, 564 
F.2d 458 (D.C. Cir. 1977); Sierra Club v. Costle, 657 F.2d 298 (D.C. Cir. 1981). 

491. See, e.g., id. The major case on the subject, of course, is Morgan v. United States, 298 
U.S. 468 (1936). See also United Steelworkers of America, et al v. Marshall, 647 F.2d 1 189 (D.C. 
Cir. 1980) (upholding OSHA's standard for occupational exposure to lead). 


The foregoing discussion ignores an important area of fruitful collaboration 
but at the same time understates the "highest common denominator" problem. 
The discussion focuses on the procedures the four agencies might follow once 
they have decided to commence regulatory proceedings. It thus ignores the earlier 
stages of agency decisionmaking in which chemicals are selected for study, their 
risks are assessed, and the costs of different controls are investigated and debated. 

The formaldehyde example provides an illustration of the opportunities for 
interagency collaboration in evaluating the toxicity of chemicals. Upon learning 
about the CUT study, the CPSC and the other IRLG agencies asked the National 
Toxicology Program to assemble a group of scientists to review the data."^^-^ The 
NT? panel conducted a careful review, concluding that formaldehyde is an animal 
carcinogen by inhalation and has not been exonerated as a human carcinogen 
by any epidemiological study yet reported. If any of the other three agencies 
should now decide to regulate formaldehyde, the panel's findings would provide 
a logical starting place. 

On the other hand, as a paradigm for coordinated rulemaking formaldehyde 
has one shortcoming. Because the agencies with jurisdiction over formaldehyde 
could choose to regulate under procedures that are in many respects similar, the 
procedural differences that may in other cases impede coordinated rulemaking 
are understated. An example that more vividly illustrates the potential obstacles 
is the food preservative, sodium nitrite. 

Nitrite is used in a wide variety of meat, poultry, and other food products 
regulated either by FDA or by USDA. While the two agencies' jurisdictions do 
not overlap functionally, the FD&C Act's food additive provisions apply to meat 
and poultry products regulated by USDA, as well as to foods that are the exclusive 
responsibility of FDA."^^^ A detailed analysis of the legal status of nitrite in these 
various products would serve no point; it is sufficient to say that nitrite's con- 
tinuing uses are subject to regulation under three different statutory provisions. "^^"^ 
The applicable administrative procedures range from informal rulemaking under 
the APA to rulemaking on a record, a process that has been criticized as among 
the most cumbersome in the administrative arsenal. "^^^ If FDA and USDA were 
to decide to end all uses of nitrite in meat and poultry products, FDA would be 
obligated to afford proponents of some uses an opportunity for a formal evi- 
dentiary hearing. Other uses, subject to different statutory provisions, could in 
theory be terminated after only the opportunity to submit written comments. The 
nitrite case is thus one in which comprehensive regulation either would follow 

492. See Merrill, Consumer Products at 1355; Risk Assessment in the Federal Government 
at 122-25. 

493. See 21 U.S.C. § 348 (1976); see generally Merrill and Hutt, Appendix A, at 948-49; 
Food Regulation: A Case Study of USDA and FDA, Chapter4of Vol. V, Study on Federal Regulation, 
Senate Comm. on Governmental Affairs, 95th Cong., 1st Sess. (1977). 

494. A summary analysis appears in General Accounting Office, Does Nitrite Cause Cancer? 
Concerns About Validity of FDA-Sponsored Study Delay Answer (1980) [hereafter GAO Nitrite 
Report] . 

495 . See generally, Hamilton, Rulemaking on a Record by the Food and Drug Administration, 
50 Tex. L. Rev. 1132 (1972); Merrill and Hutt, at 895 et seq. 


two different schedules, or the two agencies would have to adjust their common 
timetable to the requirements of the more formal procedure. 

H. Public Participation in Agency Decisions 

1 . Introduction 

This section is concerned with the opportunities for private parties to par- 
ticipate in and influence agency decisions concerning hazardous substances. The 
central question is whether agency procedures permit involvement by the several 
interests affected by their decisions, including users and distributors as well as 
persons who may be exposed. The following discussion is not intended to support 
recommendations concerning public fundings of participation in the administra- 
tive process. The issue of subsidizing participation is an important one, but it 
transcends the subject of carcinogen regulation. The purpose of this section, 
therefore, is to describe the opportunities current agency procedures afford for 
outside participation and to identify features that impede such participation. 

Broad participation in the regulatory process serves three values. Perhaps 
preeminent is the value of fairness; the "right to be heard" before government 
acts adversely to important private interests is well-established in American 
constitutional law."^^^ This right is protected by the APA and by most regulatory 
legislation. The current procedures of the agencies studied can be assumed 
constitutionally sufficient; whether they assure adequate opportunities for private 
persons to be heard in time to influence agency decisions is another question. 

A second value served by outside participation in agency decision making 
is accuracy. "^^^ In the present context this value should rank high. Decisions 
concerning human exposure to potential carcinogens affect large stakes and 
should rest on accurate and scientific judgments. Furthermore, predictions of 
risk and cost require skills that agencies sometimes lack. Accordingly, it can be 
argued, agency procedures should facilitate participation by those in the private 
sector with relevant scientific and economic expertise. One expression of this 
argument is the demand for scientific "peer review" of agency scientific judg- 
ments. "^^^ 

A third distinctive value is served by public participation in agency decisions 
concerning environmental health hazards. Such decisions take large chances with 
both human life and private business. Data are invariably inadequate and esti- 
mates of future consequences are problematical, but these uncertainties simply 
complicate an already difficult task. Determining the appropriate level of human 
exposure to a substance predicted to cause cancer is a normative exercise, in 
which there are no experts. So long as such decisions are left to regulators with 

496. See generally. Friendly, Some Kind of Hearing, 123 U. Pa. L. Rev. 1267 (1975). 

497. See Mashaw and Merrill. Introduction to the American Public Law System 325-27 

498. See generally. Risk Assessment in the Federal Government at 132-39, and sources cited 


only delphic congressional guidance, agency procedures should facilitate broad 
participation and vigorous debate /^^ 

The next four parts describe the opportunities for participation in the pro- 
grams studied for this report. While the statutory procedures for decision making 
are summarized, the focus is not on what is permitted but on what occurs: Who 
participates in the agency's process? What opportunities are afforded for access 
and when? What groups have been influential? The discussion reveals diverse 
answers to these questions. At the same time, it suggests that agency processes 
have generally been open to groups interested in making their views heard. The 
record does not illustrate a process that is closed or unresponsive to vigorous 

2 . FDA Regulation of Food Contaminants 

This report treats only one FDA program, its regulation of food contami- 
nants, which is not exemplary of the agency's general attitude toward public 
participation. Of the four agencies studied, FDA has the most elaborate proce- 
dures for structuring access to its regulatory proceedings and to its informal 
decisionmaking. It is the only one of the four agencies that initiated its own 
program for funding public participation in formal agency hearings. ^^ 

FDA's process for regulating food contaminants has not encouraged par- 
ticipation, however, either by public interest groups or by food distributors. The 
FD&C Act ostensibly requires the agency to establish formal tolerances for any 
toxic contaminant whose occurrence in food cannot be avoided. As described 
elsewhere, ^^' however, FDA has eluded this directive. Instead, it has relied on 
so-called action levels to express the amount of contamination it is prepared to 
allow before it will initiate enforcement action. Action levels are simply an- 
nounced by the agency, rather than set through proceedings which facilitate 
public participation. To be sure, FDA invites comments on the levels it announces 
and has said that it will consider evidence submitted by outside parties. In reality, 
the agency has little interest in further dialogue but, judging by the few comments 
it receives, neither do members of the public. Accordingly, FDA's action levels, 
generally determined without public participation, are the final word on most 
food contaminants. 

Several factors explain the approach FDA has devised. First, the number 
of action levels FDA must set is large, and the list is continually changing as 
new contaminants are discovered and levels of known contaminants fluctuate. 
Second, precisely because food contaminants are undesired, it is often difficult 
to find groups informed about their toxicity or interested in defending their 

499. See, e.g., Merrill and Schewel, FDA Regulation of Environmental Contaminants of 
Food, 66 Va. L. Rev. 1357, 1384-85, 1433-40 (1980). 

500. See 41 Fed. Reg. 35856 (1976) (advance notice of proposed rulemaking); 44 Fed. Reg. 
23044 (1979) (proposed rule); 44 Fed. Reg. 59174 (1979) (final regulations establishing criteria and 
procedures for financial grants to support public interest participation in formal hearings). 

501. See Merrill and Schewel. Environmental Contaminants, at 1382-91. 


presence in food. Third, FDA's desire to be able to revise these "informal 
tolerances" on short notice has made it reluctant to adopt any procedure that 
required public proceedings. Finally, the agency's reservations about the toxi- 
cological data and economic predictions underlying most of its action levels have 
discouraged it from assuming an obligation to "publish" its decisions. It has 
therefore been content with a process that does not encourage participation. ^^-^ 

A notable exception is PCB's, one of the most ubiquitous contaminants, 
whose levels in food have remained a concern to the Environmental Defense 
Fund. Over several years EDF lobbied FDA to set, and later to reduce, tolerances 
for PCB's in fish and other commodities. ^^^ This example suggests that outside 
groups can influence FDA's decisions about a contaminant if they are persistent. 
EDF has certainly been influential, perhaps instrumental, in persuading FDA to 
reduce the tolerance for PCB's in fish. The agency's recent agreement that the 
tolerance should be reduced precipitated the first demand for a formal hearing 
under section 406 — by Great Lakes fishing interests. The ongoing hearing will 
afford both groups ample opportunity to persuade FDA that its assessment of 
the health effects of PCB's and its prediction of the economic effects of a 2 ppb 
tolerance are erroneous. ^^"^ 

Distributors of food have another means to contest FDA's decisions. Since 
an action level is not a binding regulation, a distributor whose food is seized 
can challenge FDA's claim that the contaminant is likely to render the food 
injurious to health. One seafood distributor, whose shipments of mercury-con- 
taminated swordfish FDA had attempted to enjoin, did precisely this, and per- 
suaded the trial court that FDA's .5 ppm action level for mercury was 
unsupportable.^^^ Other efforts to contest FDA's action levels in enforcement 
proceedings have not generally been successful, however. ^^^ Furthermore, this 
is a mechanism available only to food distributors who believe the FDA's levels 
are too low; someone who regards them as too lenient must concentrate initially 
on the generally perfunctory administrative process. 

FDA has not sought outside help in evaluating the health effects of food 
contaminants, an approach I have recommended. ^°^ The agency appears often 
to have set its action levels to avoid destruction of substantial quantities of food. 
This may be justified by a balanced assessment of the potential health effects 
of existing contamination levels. Such an approach, however, shelters FDA's 
judgments from external scrutiny; the agency's process for setting action levels 
does not facilitate participation by outside groups, and the results do not provoke 
food distributors to resist. 

502. Id. 

503. Id. at 1416. 

504. See Food Chemical News, August 20, 1979, p. 24-25. 

505. United States v. Anderson Seafoods, Inc., 441 F. Supp. 1 151 (N.D. Fla. 1978), affd, 
622 F.2d 157 (5th cir. 1980). 

506. See, e.g., United States v. Boston Farm Center, Inc., 590 F. 2d 149 (5th Cir. 1979). 
See generally, Merrill and Hutt, at 95-106. 

507. See Merrill and Schewel, Environmental Contaminants, at 1439-40. 


3. CPSC Regulation of Chronic Hazards 

Of the four agencies studied, the CPSC appears to have been the most 
receptive to third-party influence. All of the substances the agency has attempted 
to regulate as potential human carcinogens were the subject of formal petitions 
from public interest groups. ^^^ It is a matter of debate whether any of these 
petitions was responsible for the Commission's action, but they certainly were 
influential in hastening decisions. The record is less clear concerning the CPSC's 
receptivity to arguments submitted by distributors of products considered for 

During the 1970's the CPSC conducted public proceedings on several pro- 
posals to regulate potential carcinogens. It initiated rulemaking proceedings to 
effect bans on vinyl chloride, asbestos in patching compounds and emberizing 
materials, benzene, and formaldehyde.^"^ In each instance, consumer groups as 
well as manufacturers and distributors participated by filing data and comments. 
The form of these proceedings, informal rulemaking in all cases except one, 
permits participation by any organization equipped to prepare and submit com- 
ments, and does not impose costs likely to discourage involvement by organi- 
zations with modest resources. Furthermore, the CPSC's liberal freedom of 
information policy assured public access to materials before the agency, including 
draft recommendations of staff offices.'''" 

Two of the CPSC's initiatives do not fit this description of "open decisions 
openly arrived at." In regulating TRIS, the Commission selected a legal approach 
that obviated any administrative proceedings, and ultimately remitted manufac- 
turers and distributors to challenges in court. ^" The public interest group position 
obviously was heard in this case, for EDF's petition pressured the agency to act 
when it did and outlined the legal procedure it chose. It also appears that producers 
of TRIS-treated fabric were heard informally on the scientific and economic 
issues. ^'^ However, the CPSC's approach, which it has not since repeated, 
effectively precluded dispassionate debate about the hazards of TRIS in an ad- 
ministrative forum. 

The second exception is the agency's handling of hairdryers containing 
asbestos. The CPSC resolved this dispute by negotiating agreements with man- 
ufacturers and distributors to cease distribution of the hairdryers and to repurchase 
models in consumer hands. ^'"^ It can scarcely be contended that industry interests 
were deprived of an opportunity to be heard, since any distributor could have 
demanded formal hearing under section 15.^''* And though EDF, which had 

508. See Merrill, Consumer Products, at 1363-64. 

509. See id. at 1310-46. 

510. See id. at 1370-71. 

511. See id. at 1323-32; United States v. Articles of Hazardous Substance, 588 F.2d 39 (4th 
Cir. 1978). But see Springs Mills, Inc. v. CPSC, 434 F. Supp. 416 (D.S.C. 1977) (overturning 
CPSC enforcement policy as procedurally invalid rule). 

512. See Merrill, Consumer Products, at 1325. 

513. Id. at 1349. 

514. Id. at 1285-88. 


petitioned for a ban, was not a party to the discussions between the Commission's 
staff and the distributors, its views obviously were known within the agency. If 
the Commissioners' approach harmed any interest, it was the pubHc's interest 
in learning why the risk posed by the hairdryers justified the remedy that the 
agency demanded. 

4. OSHA Regulation of Occupational Carcinogens 

If rigorous public debate is the mark of a program that has attracted the 
attention of affected interests, the Occupational Safety and Health Administration 
must be ranked a success. Its efforts to establish exposure standards for toxic 
materials have occasioned more controversy than any of the other programs 
studied for this report. Among many explanations, one is pertinent here. Unlike 
FDA, EPA, and the CPSC, whose programs are intended to protect individuals 
who have no common organizational ties, OSHA faces an organized class of 
beneficiaries, the labor unions.^ ^^ Union representatives have been active par- 
ticipants in OSHA's standard-setting activities since 1970. Indeed, OSHA was 
created to institutionalize worker protection, and its location within the Depart- 
ment of Labor contributed to the agency's concern for, and attentiveness to, the 
view of organized labor. ^'^ 

Labor unions have participated in OSHA's regulatory process in several 
ways. Most obvious, unions have submitted comments and presented testimony 
in every rulemaking proceeding to set standards for toxic materials.^' '' The AFL- 
CIO has a small staff devoted to the study of worker safety and health, and 
many individual unions have added this area to their activities.^ '^ During the 
mid-1970's, organized labor precipitated several Congressional inquiries into the 
slow pace of OSHA regulation. Unions have also sought judicial review of most 
of OSHA's health standards. In the early 1970's, the AFL-CIO was perhaps the 
sharpest critic of OSHA's standards; more recently, union challenges appear to 
have been designed primarily to influence the venue of judicial review. ^'^ 

The OSH Act affords several opportunities for participation by outside 
groups. First, it accords a special role to the National Institute of Environmental 
Health Sciences, which is supposed to provide the scientific support for OSHA's 
initiatives. ^^^ Second, section 6 of the Act specifically authorizes petitions by 
"an interested person, a representative of any organization of employers or 

515. See generally, S. Kelman, Occupational Safety and Health Administration, Ch. 7 in J. 
Wilson (ed.). The Politics of Regulation (1980). 

516. See Nichols and Zeckhauser, OSHA After a Decade: A Time for Reason, Ch. 7 in L. 
Weiss & M. Klass (eds.), Case Studies in Regulation: Revolution and Reform (1981). 

517. See OSHA Draft at 123 et seq. 

518. Interview with Grover Wrenn, former Director of Health Standards, OSHA, June 23, 

519. See, e.g.. Industrial Union Dep't v. Hodgson, 499 F.2d 467 (D.C. Cir. 1974) (challenge 
to OSHA standard for asbestos). 

520. See Risk Assessment in the Federal Government, at 109-14. 


employees [or] a nationally recognized standards-producing organization. . . ."^-^^ 
Petitions preceded several of OSHA's standards for toxic chemicals. Most have 
come from labor organizations, but OSHA's coke oven standard was a response 
to an industry petition. ^^^ OSHA's generic carcinogen policy purports to afford 
an additional opportunity for outsiders to influence its priorities by inviting 
comment on its list of candidate chemicals and its subsequent ranking of targets 
for regulation. ^^^ 

Employee organizations have also filed suits demanding that OSHA take 
action against specific hazards. The most prominent of these cases was begun 
in 1973 by the National Congress of Hispanic American Citizens, an organization 
of migratory farm workers, which sought to compel the promulgation of standards 
for the agricultural industry which OSHA had previously proposed. The plaintiffs 
argued that the OSH Act's time frames for rulemaking are mandatory, and 
triggered by any proposal. While the D.C. Circuit agreed that the Secretary of 
Labor had "discretion to alter priorities and defer action due to legitimate sta- 
tutory considerations, "^^"^ it directed the trial court to require OSHA to provide 
a timetable for development of the standards. ^^^ 

The OSH Act contemplates establishment of two types of advisory com- 
mittees on which worker and employer representatives are assured membership. 
NACOSH has recently declined in influence, but it was intended to have a role 
in determining OSHA's priorities. ^^^ Ad hoc committees to consider specific 
workplace hazards, such as coke oven emissions, initially had greater impact. 
Both labor and employer representatives considered membership to be a matter 
of real importance, not simply of appearance. Since 1976, however, OSHA has 
effectively foreclosed this avenue for participation by affected interests. ^^^ 

OSHA's rulemakings to establish standards for toxic materials have rou- 
tinely provoked an avalanche of comments and testimony from both labor and 
employer groups. ^^^ Union spokesmen have focused on health effects issues; 
they have rarely attempted to provide alternative estimates of the cost of achieving 
compliance with exposure limits, but instead attempted to deflate industry es- 
timates. Employer groups have typically addressed the full range of issues raised 

521. See 29 U.S.C. § 655(b)(1) (1976). 

522. See 41 Fed. Reg. 46742 (1976). The industry petitioners actually sought a relaxation of 
the existing standard adopted by OSHA immediately following enactment of the Occupational Safety 
and Health Act. 

523. OSHA Carcinogen Policy, 45 Fed. Reg. 5001, 5207-08 (1980). 

524. National Congress of Hispanic-American Citizens v. Usery, 554 F.2d 1196 (D.C. Cir. 
1977). See also 425 F. Supp. 900 (D.D.C. 1975). 

525. Id. at 1200. 

526. Interview with Charles Gordon, Esq. , Office of the Solicitor, Department of Labor, June 
23, 1981. 

527. Id. See also Wrenn Interview. 

528. See Risk Assessment in the Federal Government at 94-99; McGarity, The Occupational 
Safety and Health Administration's Generic Carcinogen Policy: Rulemaking Under Scientific and 
Legal Uncertainty (unpublished paper prepared for the National Center for Administrative Justice, 
April 1981). 


by OSHA's proposals. The agency has also frequently initiated the standard- 
setting process by publishing an advanced notice of proposed rulemaking; this 
technique, mandated by the agency's carcinogen policy, affords an additional 
opportunity for outside interests to be heard. ^^^ Furthermore, judicial review of 
OSHA's health standards has been the rule; acquiescence has more often signalled 
surrender than agreement. 

OSHA has never provided funding for participants in its rulemaking process. 
Most of OSHA's health standards have applied to industries in which employees 
are well organized, and employers have been vigorously opposed. One former 
OSHA official expressed the view that the union representative sometimes lacked 
knowledge and sophistication. '''^^ This view was not, however, echoed by other 
observers. ^^' OSHA's recent rulemaking brought forth as many as 100 witnesses 
and routinely produced enormous records. ^^^ While perhaps poor proxies of the 
effectiveness of participation, these figures do not suggest a process to which 
only one side has access. 

Assessment of the impact of union and employer participation on OSHA's 
decisions is difficult. Different interests have been influential, but one cannot 
attribute their success to any particular procedural device. OSHA has modified 
proposed exposure limits in response to submissions from both employers and 
employee organizations. Their contrasting positions usually bracket OSHA's 
final choices; to the extent that they balance each other, both may be influen- 

In one respect, parity does not exist between unions and employers in the 
OSHA standard-setting process. Toxicological data are usually equally accessible 
to all participants. Studies of exposed workers, though based largely on employer- 
generated data, are increasingly well-publicized. But employers are thought to 
have a significant edge in their access to information about the feasibility of 
compliance with alternative exposure levels. Information about alternative meth- 
ods of achieving exposure targets and their likely cost is closely held and re- 
luctantly shared. ^^"^ 

529. See, e.g., 37 Fed. Reg. 13285 (1972) (NIOSH request for data on fourteen potential 
carcinogens); 39 Fed. Reg. 44769 (1974) (ANPRM for cotton dust standard); 42 Fed. Reg. 33043 
(1977) (request for data on acrylonitrile). 

530. Interview with Anson Keller, Esq. , former Executive Assistant to the Assistant Secretary 
of Labor for Occupational Safety and Health, August 1980. 

531. E.g., Interview with Robert Barnard, Esq. counsel to the American Industrial Health 
Council, August 1980; Wrenn Interview, supra. 

532. See McGarity, note 528 supra, OSHA Draft at 124-25 and Table I. 

533. See, e.g., the discussion of union and employer submissions in American Textile Mfgrs. 
Institute v. Donovan, 101 Sup. Ct. 2478 (1981) (cotton dust standard); United Steelworkers of 
America, et al. v. Marshall, 641 F.2d 1 189 (D.C. Cir. 1980) (lead standard). See also OSHA Draft 
at 124 et seq. 

534. See, e.g., Wrenn Interview; American Textile Mfgrs. Institute v. Donovan, 101 Sup. 
Ct. 2478 (1981) (accepting OSHA's estimate of compliance costs in light of agency's asserted 
inability to obtain detailed confidential data from industry). 


5. EPA Regulation of Pesticides 

The influence of outside groups on the Environmental Protection Agency's 
regulation of pesticides has fluctuated over the last decade. Originally excluded 
from the agency's deliberations, such organizations as EDF became important 
participants in EPA proceedings during the early 1970's. As the program has 
evolved in the past several years, however, their activities have diminished. In 
this same period, the role of academic scientists and agricultural interests has 

FIFRA is the one statute among those examined that creates legally protected 
interests in the production and distribution of toxic chemicals. As a licensing 
law it accords distinctive procedural rights to registrants and applicants for reg- 
istration, i.e., manufacturers and distributors of pesticides. ^^^ Among other pro- 
tections, they are assured an opportunity for a trial-type hearing on EPA's refusal 
to register a pesticide or on its decision to cancel registration. Pesticide producers 
have substantial opportunities to participate in EPA's decisionmaking process; 
indeed, they historically have supplied most of the information on health effects 
and on commercial utility on which the agency has relied. Over the past decade, 
the issue has been whether the system affords adequate opportunities for other 
interests — including users, environmental groups, and health scientists — to con- 
tribute to EPA's decisions. 

In the late 1960's, environmental groups found the system for pesticide 
regulation — then administered by USDA — resistant to growing concerns about 
the environmental effects of many pest control agents. ^^^ The process for reg- 
istering new pesticides had become a private dialogue between applicants and 
the agency. More troubling, for registered pesticides the law afforded no ready 
forum for public ventilation of safety questions. Between 1969 and 1972, EDF 
won several court victories over USDA and EPA which opened the process to 
public scrutiny. Two major cases established the propositions, first, that EPA 
could not indefinitely delay a response to petitions claiming that a pesticide posed 
an "imminent hazard" to health^"^^ and, second, that the EPA was obligated to 
initiate cancellation hearings for any pesticide about which "a substantial ques- 
tion of safety" was raised. ^-^^ One result of these rulings was to force EPA to 
commence the cancellation process whenever it received data indicating that a 
pesticide might cause cancer. While environmental groups had no clear claim 
to a hearing in their own right, FIFRA did allow them to participate in any 
hearing requested by a manufacturer whose registration was at issue. EDF par- 
ticipated actively in EPA's later hearings on such agents, as DDT, aldrin and 
dieldrin, and heptachlor and chlordane.^^^ 

535. Seel U.S.C. § 136d (1978). 

536. See, e.g., Environmental Defense Fund, Inc. v. Hardin, 428 P. 2d 1093 (D.C. Cir. 
1970); Environmental Defense Fund, Inc. v. Ruckelshaus, 439 F.2d 584 (D.C. Cir. 1971). See 
generally, EDF and R. Boyle, Malignant Neglect 117-36 (1979). 

537. EDF V. Hardin, 428 F. 2d 1093 (D.C. Cir. 1970). 

538. EDF V. Ruckelshaus, 439 F.2d 584 (D.C. Cir. 1971). 

539. See EDF and R. Boyle, supra, note 536. 


In 1975 and 1976, Congress and EPA each modified the system for regu- 
lating problem pesticides. FIFRA was amended to accord greater procedural 
protection to registrants and to require that EPA consult both with USDA and 
with independent scientists — the Science Advisory Panel — before initiating can- 
cellation proceedings. ^^^ At approximately the same time, EPA created the "re- 
buttable presumption against registration" (RPAR) process, which responded to 
the D.C. Circuit's demand that the agency open up its internal review of pesticides 
considered for cancellation and simultaneously subordinated the role of formal 

According to EPA, deemphasis of adjudicatory hearings reflected a desire 
to permit broader public participation in its decisions about pesticides. ^"^^ Hear- 
ings were thought to be expensive and time-consuming and their adversarial 
character offended members of the scientific community. Furthermore, since by 
statute a hearing followed publication of a notice to cancel registration, the public 
was given the "impression that the Agency had made up its mind already" about 
the fate of the pesticide. ^"^-^ The RPAR process introduced an opportunity for 
informal public debate over the risks and benefits of a pesticide before EPA 
initiated cancellation proceedings. 

It is unclear, however, whether the RPAR process has stimulated public 
participation. The agency did not invite public comment on its initial list of 
RPAR candidates. ^'^'^ On the other hand, its regulations set low thresholds for 
issuance of a rebuttable presumption, thus triggering public discussion of the 
risks and benefits of more products. ^"^^ Participation in the RPAR process is 
clearly less costly than formal appearance in an adjudicatory hearing, and does 
not preclude intervention in any hearing that is held. Furthermore, the amend- 
ments to FIFRA not only forced EPA to confer with USDA and SAP, they 
created — in SAP — an additional forum in which both registrants and opponents 
are able to contest the agency's assessment of risk.^"^^ In sum, current EPA 
procedures for evaluating registered pesticides afford numerous opportunities for 
public participation. Whether interested groups have the resources to participate 
effectively is, obviously, a separate question. Environmental groups are probably 
less involved now in pesticide regulation than in the early 1970's. But their 
diminished interest may have other causes, including preoccupation with other 
EPA programs (in particular with TSCA), reduced visibility of the pesticide 
program, and the increased complexity of the issues involved. 

540. See text accompanying notes 281-83, supra. 

541. See 40 Fed. Reg. 28242 (1975). 

542. Id. See also Pesticide Programs; Rules Governing Rebuttable Presumption Against Reg- 
istration (RPAR) Proceedings; Rules of Practice Governing Hearings Under Section 6 of the Federal 
Insecticide, Fungicide, and Rodenticide Act (FIFRA), 45 Fed. Reg. 52628 (1980) (proposed revisions 
of RPAR procedures). 

543. 40 Fed. Reg. 28242 (1975). See Interview with Steven Jellinek, Former Assistant Ad- 
ministrator for Pesticides and Toxic Substances, EPA (June 23, 1981). 

544. Id. 

545. Id. See also 45 Fed. Reg. 52628 (1980). 

546. See text accompanying note 281 et seq, supra. 


There are, still, two legal barriers to full participation by non-registrants. 
Both stem from the language in FIFRA that assures registrants an evidentiary 
hearing on cancellation of pesticide registrations. ^"^^ According to the D.C. Cir- 
cuit, this language does not obligate EPA to hold a hearing on an environmental 
group's allegations that the agency improperly failed to cancel all unsafe uses 
of a pesticide. ^"^^ Following issuance of a cancellation notice for chlorobenzilate, 
EPA and the registrants agree to cancellation of most uses but continued reg- 
istration for Hawaiian pineapples. After the registrants withdrew their hearing 
request EPA denied EDF's demand to be allowed to demonstrate that the hazards 
of chlorobenzilate outweighed the benefits of that remaining use as well. The 
Court of Appeals agreed that FIFRA did not accord EDF a right to a hearing on 
whether EPA's initial action adequately protected health and the environment. 
While this ruling does not diminish opportunities for third party participation in 
the pre-cancellation RPAR process, it allows EPA to avoid formal hearings on 
any decisions that do not precipitate hearing demands from pesticide registrants. 

A second ruling similarly forecloses hearings at the insistence of pesticide 
users who wish to defend a pesticide in the face of acquiescence by its registrants 
to cancellation. In McGill v. Environmental Protection Agency, ^^"^ farmers, op- 
erators of recreational facilities, and local governmental organizations in Mis- 
sissippi unsuccessfully attempted to force EPA to hold a hearing on the cancellation 
of Mirex, a pesticide used to control fire ants. The Fifth Circuit held that FIFRA 
accorded hearing rights only to registrants or applicants for registration. The 
petitioners could obtain a formal hearing to explore the risks and benefits of 
Mirex only by applying for registration themselves. 

These two rulings reflect EPA's persistent efforts to avoid formal hearings 
and focus public debate over controversial pesticides in the more informal RPAR 
process. That process affords members of the public, including registrants, at 
least two opportunities to submit evidence and arguments. ^^^ Because it lacks 
an oral hearing component, however, it does not allow presentation of live 
witnesses or cross-examination. Other laws under which carcinogens are regu- 
lated, such as the CPS Act and the OSH Act, are similarly deficient, although 
both provide informal oral hearings and OSHA has voluntarily accorded rights 
of cross-examination. Accordingly, the adequacy of EPA's process for regulating 
problem pesticides comes down to the question whether trial-type procedures 
are essential for resolving the sorts of issues these decisions pose. 

The current FIFRA procedures do not accord equal access to all interested 
persons. The statute assures formal hearings to registrants who desire to contest 
cancellation; it merely permits nonregistrants to participate in hearings that are 
held. Registrants are thus able to control the type of procedure EPA will use in 

547. 7 U.S.C. § 136d(d) (1978). 

548. See Environmental Defense Fund\. Costle, 631 F.2d 922 (D.C. Cir. 1980). 

549. 593 F.2d 631 (5th Cir. 1979). 

550. See discussion of the RPAR process in 45 Fed. Reg. 52628 (1980). 


disputed cases. This preferential treatment for licensees is not novel, ^^' of course, 
but arguably it inhibits full ventilation of the health and economic issues posed 
by pest control agents. It should be noted, however, that FIFRA does not oth- 
erwise accord registrants special rights. In the RPAR process, all persons stand 
on essentially the same footing, and if a formal hearing is held, environmental 
groups and pesticide users can participate on essentially the same footing with 

6. Participation in Informal Agency Decisions 

Although it did not arise in one of the programs studied for this report, a 
recent episode involving EPA illustrates another dimension of the public partic- 
ipation issue. The episode concerns formaldehyde, which two recent bioassays 
have shown causes cancer in rodents exposed by inhalation. The results of the 
first of these studies prompted the CPSC to propose a ban on UF foam insula- 
tion. ^^^ The five IRLG agencies had previously selected formaldehyde as one 
of six chemicals to be evaluated and possibly regulated jointly. Anticipating that 
formaldehyde might be a candidate for regulation under TSCA, EPA's staff 
undertook to evaluate both the positive animal data and the human studies that 
found no effects from exposure to the chemical. It also sought to determine the 
levels and duration of human exposure in settings not regulated by other agencies. 

In early 1981 EPA received a petition from the National Resources Defense 
Council (NRDC) asking that it commence proceedings to regulate formaldehyde 
under section 4(f) of TSCA. ''^'^ That provision obligates EPA to launch a formal 
process to consider regulation of any chemical that new information suggests 
may pose a ''substantial risk."^^**"^ The Act also requires the agency to respond 
to any petition within 90 days. NRDC claimed that the evidence of formalde- 
hyde's carcinogenicity met the requirements of section 4(f) and threatened to 
sue EPA if it failed to act. 

The merits of NRDC's claim are not critical in this context. What is of 
interest is the process by which EPA proceeded to evaluate formaldehyde to 
determine whether section 4(f) should be invoked. The statute says nothing about 
how this shall occur. Presumably it would be lawful for EPA to conduct its 
investigation relying wholly on its own staff and permitting outside participation. 
The agency's actual deliberations undoubtedly involved numerous written ex- 
changes and meetings among staff scientists and others responsible for advising 
the Administrator. They also, however, included several private discussions with 
one group of interested parties. 

In early 1981 the trade press reported that EPA's Office of Pesticides and 
Toxic Substances had recommended that the Administrator initiate proceedings 

55 L The hearing provisions of the Federal Food, Drug, and Cosmetic Act applicable to 
withdrawal of approval for new drugs similarly accord rights only to license holders. See 21 U.S.C. 
§ 355(d)-(e) (1976). 

552. 46 Fed. Reg. 11188 (1981); see Merrill, Consumer Products at 1357-60. 

553. 15 U.S.C. § 2603(f) (1976). 

554. Id. 


under section 4(f) with respect to formaldehyde. ^^^ By September, however, it 
had been reported that the new Assistant Administrator had countermanded this 
advice and recommended no action at the present time. Thereafter, as Congres- 
sional oversight hearings later disclosed, several meetings were held between 
EPA's Deputy Administrator and other agency officials, and representatives of 
the Formaldehyde Institute and the Chemical Manufacturers Association. ^^^ 
Questioning of EPA officials revealed that the meetings dealt primarily with the 
interpretation of the health effects data on formaldehyde, but they inevitably also 
touched on whether EPA should act under TSCA. It is a matter of some dispute 
who was responsible for organizing the meetings. The private organizations may 
have initiated the dialogue as they sought to persuade EPA officials not to act 
against formaldehyde, but it also appears that Deputy EPA Administrator Her- 
nandez scheduled and set the agenda for some of the sessions. ^^^ 

The importance of these private meetings stems from the fact that EPA 
eventually decided not to initiate regulation of formaldehyde under section 4(f).^^^ 
Though not formally announced until February 1982, this decision has been 
officially ascribed to the previous September, before the meetings took place. ^^^ 
NRDC has threatened to seek judicial review of EPA's failure to act. If it does 
so, it will proceed without having had an opportunity to know about or respond 
to the contentions made by users and manufacturers of formaldehyde. 

The process EPA followed in reaching its decision on formaldehyde has 
been sharply criticized. EPA Administrator Gorsuch has acknowledged that the 
agency's approach reflected insensitivity to appearances. ^^° Others have con- 
tended that EPA's conduct, which accorded preferential access to formaldehyde 
users and manufacturers, was unlawful. These claims smack of overstatement, 
but they cannot be discounted too quickly. One asserted ground of illegality is 
EPA's supposed violation of the prohibition against ex parte contacts. While 
EPA might have been constrained to avoid ex parte communications once it 
initiated rulemaking under section 4(f), Home Box Office is by its own terms 
limited to post-proposal communications. ^^' An argument can be made, however, 
that section 4(f) presents a special case. EPA's decision not to regulate is final; 
it had a statutory responsibility to reach some decision within a specific period, ^^^ 
and Congress has explicitly authorized judicial review of decisions not to initiate 

555. See generally, T. Bartman, Informal Meetings Between Agencies and Outside Groups 
(unpublished manuscript based on interviews and EPA documents obtained pursuant to the Freedom 
of Information Act, May 17, 1982). 

556. 9 Pesticide and Toxic Chemical News 6 (No. 49, 1981). 

557. See Bartman, supra, at 4; Letter from Dr. John Hernandez, Deputy Administrator, to 
Congressman Toby Moffett, October 5, 1981. 

558. Memorandum from Dr. John A. Todhunter, Assistant Administrator for Pesticides and 
Toxic Substances, to Anne Gorsuch, EPA Administrator, February 10, 1982. 

559. See Bartman, supra, at 3. 

560. 9 Pesticide and Toxic Chemical News 6, No. 49 (1981). 

561. Home Box Office, Inc. v. Federal Communications Comm'n, 567 F.2d 9 (D.C. Cir. 

562. TSCA § 4(f), 15 U.S.C. § 2603(f) (1976). 


rulemaking under section 4(f).^^^ It is not far-fetched to argue that ex parte 
communications should be avoided, or at least documented and disclosed, in 
this context as well. 

EPA's second asserted breach of mandated procedures in its handling of 
formaldehyde is grounded on the Federal Advisory Committee Act (FACA),^^"^ 
which governs outside groups from which an agency seeks advice. The FACA 
requires formal chartering, advance notice of meetings, and, with exceptions not 
relevant here, public access to all committee sessions. None of these formalities 
was complied with by EPA. The courts have held the FACA applicable to many 
informal efforts by agencies to consult with private groups. ^^^ The D.C. Circuit 
has also made clear that it does not apply to meetings held at the request of 
private groups to make their position known to agency officials. ^^^ The historical 
record is ambiguous, concerning the initiative for and content of the meetings 
Dr. Hernandez held with the Formaldehyde Institute and CM A. 

Even assuming that EPA officials violated no law in meeting privately with 
representatives of firms which made and used a substance under consideration 
for regulation, their conduct is still open to question. It leaves the distinct 
impression that EPA's decision not to regulate formaldehyde was the product 
of proceedings in which other interests had no opportunity to participate. EPA's 
failure for several weeks to provide any statement of reasons for its decision 
accentuated this impression. Equality of access to agency decisionmakers and 
the appearance of administrative integrity both warrant changes in EPA's ap- 

The formaldehyde episode has not been selected to criticize EPA's new 
management. The kind of decision that EPA confronted arises frequently in 
agencies responsible for regulating environmental hazards: A new study, sug- 
gesting hazards previously unsuspected, precipitates internal review by agency 
scientists. If the study findings are reliable and disturbing, regulatory action may 
be appropriate; if they are not trustworthy or they are contradicted by better 
evidence, probably no action is required. An agency cannot follow a formal, 
highly structured process in making all such decisions, for they arise too fre- 
quently. ^^^ It must be able to rely on internal resources in deciding what issues 
are important enough to set in motion the statutory machinery for decisionmaking. 

Furthermore, officials of an agency should not be precluded from meeting 
with persons interested in its decisions. Whether such meetings should always 
be public is debatable, although I think the better view is that any individual 

563. Id. 

564. Pub. L. No. 92-463, codified at 5 U.S. C. Appendix I (1976), as amended by Pub. L. 
No. 94-409, 90 Stat. 1247 (1976). 

565. E.g., Food Chemical News v. Davis, 378 F.Supp. 1048 (D.D.C. 1974). 

566. Consumers Union v. Department of HEW, 409 F.Supp. 473 (D.D.C. 1967), affd without 
opinion, 551 F.2d 466 (D.C. Cir. 1977). 

567. Regulatory agencies are constantly receiving reports of toxicological investigations, 
conducted by industry or reported in the scientific literature, about substances that potentially are 
subjects of regulation. 


should be entitled to meeting privately with government officials. Recognition 
of such a "right" need not prejudice the public interest in knowing what officials 
are hearing, nor impair the opportunity of persons with opposing views to be 
heard. Two safeguards would protect both interests. First, an agency decision- 
maker who agrees to meet privately with parties to a controversy should be 
willing to meet on the same terms with opposing parties. Second, the agency 
should insist that all written submissions by any party be available to the public 
and the decisionmaker should prepare, and release, a memorandum summarizing 
arguments made orally. This approach has been followed by FDA since the mid- 

To be sure, this proposal goes beyond the specific requirements of the Home 
Box Office decision, for it would apply whether or not the agency had published 
a notice of proposed rulemaking. But the formaldehyde episode demonstrates 
that important issues are often resolved, and major interests affected, by decisions 
that do not result in public proceedings. Indeed, decisions by agencies not to 
initiate regulation of potentially hazardous chemicals may be as significant — 
for both human health and economic welfare — as decisions reached after formal 
public proceedings in which all affected interests have participated. 

568. FDA practice, for example, assures any citizen of the opportunity to request and obtain 
a private meeting with agency officials. A memorandum of any such meeting will be prepared by 
the agency officials present and made available for public inspection, with any trade secret or other 
legitimately confidential information deleted. High-ranking agency officials are required to maintain 
and disclose a calendar of all such meetings, including telephonic meetings, so that members of the 
public can be aware of their occurrence. See 21 C.F.R. Part B. 


Thomas J. Madden and Nicholas W. Allard"^ 



I. Federal 
II. State 

I. Introduction 
The Subject 

The AC US Mandate 

II. Analytic framework 
Key Concepts 

Competing Objectives, Core Values 

Alternative Methods of Dealing With Governmental Wrongs 

III. The Existing System of Dealing with Governmental Wrongs 

A. The Private Damage Action 

1 . Expansion of Individual Liability For Money Damages 
— Judicially Discovered Rights — General Trends 

— Section 1983 Actions 
— Bivens Actions 
— Litigation Costs — 

Attorneys' Fees 
— Litigation Costs — 

Government's Defense of Officials 

2. The Immunity Doctrine 

— Evolution of the Case Law 

— The Case Law's Matrix of Issues 

a. Source of Immunity 

b. Degree of Immunity Available 

c. Discretionary and Mandatory Action 

d. Scope of the Defendant's Authority 

e. Piercing the Immunity Defense 
— An Evaluation 

B. Existing Forms of Public Enterprise Liability 
1. Federal Enterprise Liability 

— Insurance 

* Thomas J. Madden and Nicholas W. Allard are attorneys at Kaye, Scholer, Fierman, Hays 
& Handler, Washington, D.C. This report was prepared with the assistance of Patrick R. Harkins 
and David H. Remes, also of that firm. 



— Indemnification 

— Direct Government or Enterprise Liability 
2. State Enterprise Liability 

C. Administrative Controls Over Federal Ojficials 
1. Individual Sanctions 

— Elected Officials 
— Political Appointees 
— Excepted Service 
— Competitive Service 

a. Reduction in Grade or Removal by the Agency for Unac- 
ceptable Performance 

b. Disciplinary Sanctions Imposed By the Agency to "Promote 
the Efficiency of the Service" 

D. External Controls 

1 . Impeachment 

2. The Electorate 

3. Congressional Control 

4. Special Prosecutor 

5. Public Pressure 

The role of Administrative and External Controls in the Civil Sanction 

IV. Analysis: Changes at the Federal Level 

A . Summary of Issues Raised by Proposals to Establish Federal Enter- 
prise Liability 

1 . Proposed Legislation 

a. H.R. 7034 

b. S. 1775 

2. Assessment of Issues Raised by H.R. 7034 and S. 1775 

a. Retention of the "Good Faith" Defense 

b. The Search for an Effective Substitute Deterrent 

c. Jury Trial and "Additional" Damages 

d. Attorney Fees 

e. Torts of Former Officials 

f. "Scope of Office" versus "Color of Law" 

g. The Problematic Distinction Between Constitutional and Non- 
Constitutional Torts 

B . Recommendation 

V. Analysis: Changes at the State and Local Level 

A. Liability Under 42 U.S.C. § 1983 

1 . Reprise 

2. Problems Related to Federal Assistance 

a. Rights and Obligations Under Section 1983 

b. Remedies Under Section 1983 

c. Increased Litigation 

B. State Responses to Section 1983 Liability 



C. The Congressional Response to Section 1983 Liability: Legislative 

D. A New Legislative Option 

1 . Exhaustion 

2. Exceptions to Exhaustion 

3. Minimum Standards 

E. Recommendation 
Volume II: Appendices** 

Appendix A: Statutory^ Materials 

Appendix B: State Survey 

Appendix C: Annotated Survey of the Literature 

** Some of the appendices for this report have not been pubHshed herein. They are the 
complete Appendix A: Statutory Materials and Appendix B: State Survey. These documents are 
stored in the project files of the Administrative Conference of the United States. 




This study addresses the problems surrounding current rules governing of- 
ficial liability and immunity and evaluates alternative public policy responses to 
these problems. The focus of this study is the increasing vulnerability of executive 
branch officials at the federal, state and local levels to personal liability for 
money damages. The study concludes that the existing liability and immunity 
rules result in a civil sanction system that is not optimal, and recommendations 
are offered on desirable changes. 

The recommendations of this study are based on a public policy analysis 
of official liability and immunity rules. The purpose is not to give advice on 
appropriate action under existing law or even to predict the course of legal 
developments. Rather, the objective is to identify the framework of the policy 
issues in this field and then to offer advice on changes that are worthwhile and 
feasible. The study also appraises the arguments commonly voiced in the ongoing 
debate over the subject of lawsuits against public officials. It examines relatively 
unexplored but important areas such as the role of administrative controls in a 
civil sanction system and the problem of dealing with governmental wrongs at 
the state and local levels. It is also intended that the comprehensive nature of 
this study, and the extensive materials compiled in the Appendices, will provide 
a useful tool in the continuing public examination of existing rules for official 
liability and immunity. 

Although recognizing that existing liability and immunity rules have been 
developed and will inevitably continue to be significantly influenced by the 
courts, the study recommends a legislative response to the problems arising from 
governmental wrongs. Only legislation can produce a comprehensive and co- 
herent change that addresses the complex legal, administrative, and social aspects 
of the problems. Legislation is also appropriate because of the need to accom- 
modate the competing interests such as the desire for governmental accounta- 
bility, the need to motivate government officials to engage in socially beneficial 
conduct, and the interest in minimizing government costs and inappropriate 
burdens on government officials. Moreover, the widespread interest in pending 
legislation before Congress and a significant amount of new state legislation 
demonstrate that the legislative forum will maximize public participation in 
revising the present civil sanction system. 


/. Federal 

Under current law, individual federal officials may be held personally liable 
for constitutional violations they are found to have committed while acting within 
the scope of their office or employment. Damages may not be recovered against 
the United States for violations of constitutional rights as such, although claims 


arising out of the same conduct may sometimes be stated against the United 
States under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680. 
There is nearly universal agreement that the existing system of civil sanctions 
for constitutional violations by federal officials neither provides adequate as- 
surance of compensation for victims of such violations, nor affords the degree 
of measured deterrence required to discourage improper conduct by government 
officials without discouraging proper conduct as well. In addition, the federal 
government often has interests at stake in constitutional tort litigation involving 
its officials which cannot adequately be represented by the individual officials 
themselves as defendants. 


To serve the primary goals of compensation, deterrence, and fairness in 
dealing with constitutional torts committed by federal officials, and to afford a 
solution to the problems perceived to flow from the current system of individual 
liability, Congress should enact legislation providing that the United States shall 
be exclusively liable for damages for torts arising under the Constitution of the 
United States and committed by federal officials while acting within the scope 
of their office or employment. 

Such legislation should provide: 

(1) That, in constitutional tort actions, the United States may not assert as 
a defense the absolute or qualified immunity of the official whose conduct gave 
rise to the claim, or his reasonable good-faith belief in the lawfulness of his 
conduct. Such immunities, and the good-faith defense, have been judicially 
created for policy reasons to assure that exposure to personal liability shall not 
deter individual officials from the vigorous discharge of their responsibilities, 
and serve no purpose when the government is substituted as defendant other 
than to limit the extent to which genuine victims of constitutional injuries may 
secure redress. In providing that the United States shall not assert such immunities 
or the good-faith defense. Congress may wish to provide that the United States 
shall be permitted to assert any applicable immunities of the President, and, 
perhaps, of federal judges and members of Congress. 

(2) That there shall be vested in an office of the Executive Branch inde- 
pendent authority to investigate constitutional tort cases in which there has been 
a judgment of liability against or money compromise by the United States, and 
that such office shall be vested with independent authority to conduct disciplinary 
proceedings in such cases as may be appropriate. Such a disciplinary mechanism 
is essential to perform the deterrent and corrective functions now served by the 
damage action remedy against individual government officials; existing admin- 
istrative disciplinary mechanisms, by themselves, are neither directed toward 
nor capable of performing these functions under current law. 

(a) One approach for implementing this recommendation would be to 
establish an Office of Disciplinary Counsel to investigate and, where appropriate, 
to prosecute official misconduct before an administrative tribunal independent 
of the agency that employed the offending official. Another approach would be 


to augment the existing authority and responsibilities of the Office of Special 
Counsel and the Merit Systems Protection Board under 5 U.S.C. §§ 1201-1209 
to perform the functions described herein. 

(b) If Congress should choose to rely on the Office of Special Counsel 
and the Merit Systems Protection Board to implement this recommendation, 
separate mechanisms would nevertheless be required to provide for independent 
investigation and disciplinary proceedings with respect to federal officials not 
subject to the jurisdiction of the Office of Special Counsel and the Merit Systems 
Protection Board. 

(c) Notwithstanding the authority of the independent disciplinary mech- 
anism utilized to implement this recommendation, the agency that employed the 
offending official would be responsible in the first instance for investigation and, 
where appropriate, for disciplining the official or implementing other corrective 
steps, and such action should not be postponed pending the outcome of any 
constitutional tort suit that may be filed against the United States. Congress 
should provide every federal agency with explicit statutory authority to employ 
existing administrative mechanisms for disciplining officials found to have vi- 
olated constitutional rights. 

(d) Before proceeding with disciplinary action against any official for a 
violation of constitutional rights, the Disciplinary Counsel or Special Counsel 
should determine whether the agency that employed the official has already taken 
disciplinary or other corrective action, and, if so, whether the agency's action 
is satisfactory from the standpoint of the system's overall goals of deterrence 
and accountability. The Disciplinary Counsel or Special Counsel would be au- 
thorized to initiate disciplinary proceedings before the Merit Systems Protection 
Board or other independent administrative tribunal only if he determined that 
the agency that employed the official had taken no disciplinary or other corrective 
action, or that the agency's action was insufficient. Any discipline later imposed 
by the Merit Systems Protection Board or other independent administrative tri- 
bunal would be reduced by any discipline imposed by the agency in its own 

(e) The Disciplinary Counsel or Special Counsel should be directed to 
compile and forward to the House and Senate Judiciary Committees periodic 
reports setting forth the cases he has reviewed, and explaining with particularity 
why he chose to proceed or not to proceed with independent disciplinary action 
in each case. 

(3) That Congress should provide not only for actual damages but for 
reasonable liquidated damages in the event that actual damages are nominal or 
nonexistent because the injury caused by the violation of a constitutional right 
is of an intangible nature. Congress should also consider allowing "additional" 
damages against the United States in cases where the conduct giving rise to the 
tort was undertaken with the malicious intention to cause a deprivation of con- 
stitutional rights or with reckless disregard for the plaintiff's constitutional rights. 

(4) That the right to jury trial should be retained for plaintiffs whose 
claims had arisen as of the effective date of the legislation implementing this 


recommendation, and that Congress should consider extending the jury trial right 
to plaintiffs whose claims arose subsequent to the effective date of the legislation. 

(5) That Congress consider the appropriateness of allowing attorney fees 
under such legislation in the context of a more comprehensive review of the 
attorney fee issue in across-the-board federal legislation. 

(6) That legislation implementing this recommendation provide a mech- 
anism for holding former government officials accountable for constitutional 
torts committed while they were in office, where a plaintiff has secured a judg- 
ment against or money compromise by the United States. Congress should con- 
sider allowing federal officials leaving government service to elect either to be 
sued individually for such torts, or to have the United States substituted in their 
place; and, in electing the latter, such officials would agree to submit to appro- 
priate sanctions imposed by the agency for which they were employed when 
their tortious conduct occurred. 

(7) That such legislation should provide for periodic review by the House 
and Senate Judiciary Committees of the operation of the legislation implementing 
the new system of federal enterprise liability and independent administrative 

//. State 

The Supreme Court's expansion of the scope of 42 U.S. C. § 1983 to include 
claims based on alleged violations of federal statutory laws gives rise to the 
spectre of an increase in the number of section 1983 actions against state and 
local officials and municipalities involved with federal assistance programs. The 
continued devolution of responsibility for managing federal programs to the state 
and local levels is likely to add further to the rise in these types of section 1983 

Although most states have some system to insulate government officials 
from personal liability under section 1983, increased litigation under that section 
is sure to have an adverse impact on state and local administration of federal 
assistance programs. To date, efforts to address the problem of expanded section 
1983 liability have focused exclusively on amending section 1983 by deleting 
the phrase ''and laws," thereby negating the effect of the Supreme Court's 
decision in Maine v. Thiboutot. 

However, in light of the importance of section 1983 in the vindication of 
personal rights, amending that section is the least viable alternative for amelio- 
rating the somewhat more discrete problem of derivative liability and increased 
litigation involving federal assistance programs. Instead, following the Supreme 
Court's decision in Patsy v. Board of Regents of Florida and the suggestions 
made by several of the Justices in that case, this recommendation calls upon 
Congress to consider legislation that would require the exhaustion of state ad- 
ministrative remedies as a prerequisite to bringing an action under 42 U.S.C. § 
1983 for an alleged violation of the terms, conditions or rights created by a 
federal assistance program by any state or local official or political subdivision 
of a state. 


The purpose of this recommendation is to avoid an increase in the number 
of claims hkely to be filed under 42 U.S.C. § 1983 against state and local 
officials and municipalities involved in the administration of federal assistance 
programs, and to provide an alternative system for the resolution of such claims. 
In addition, this recommendation seeks to preserve existing remedies for the 
vindication of federal constitutional rights; provide an adequate, less costly and 
less time-consuming process for resolving disputes that arise out of federal 
assistance programs; encourage state and local governments to maximize man- 
agerial efficiency of programs and projects funded by federal assistance; ame- 
liorate the adverse impact that section 1983 litigation would have on the successful 
operation of federal assistance programs by state and local entities and officials; 
and reduce the level of intervention by the federal judiciary in the administration 
of federal assistance programs. The recommendation set forth below should thus 
be applicable to all federal assistance programs benefiting state and local entities 
and third parties. Legislation to require exhaustion of state administrative rem- 
edies should contain the following general provisions: 

(1) A third party or intended beneficiary of a federal assistance program 
would be required, prior to bringing an action under section 1983, to exhaust 
state administrative remedies which have been determined to be in substantial 
compliance with minimum standards promulgated by the federal agency having 
primary responsibility for the assistance program which forms the basis for the 
third party complaint or grievance. 

State receipt of federal assistance would not be conditioned upon that state's 
agreeing to establish administrative remedies in compliance with minimum fed- 
eral standards. Instead, state provision of adequate administrative remedies would 
be voluntary. However, the exhaustion requirement would only be applicable in 
those states where such remedies have been implemented. 

(2) Claims based purely on alleged federal constitutional violations would 
not be subject to exhaustion in any case. In addition, a federal court would be 
authorized to grant interim relief in appropriate cases in accordance with federal 
rules of civil procedure, and would also be authorized to waive the exhaustion 
requirement in any case in which the court determines that exhaustion would be 
futile or the administrative remedy would be clearly inadequate. 

(3) A federal agency having primary responsibility for the administration 
of federal assistance programs would promulgate — in consultation with state and 
local entities and interested persons and groups — minimum standards to guide 
states in establishing administrative procedures and remedies for third-party com- 
plaints relating to such programs. Such standards would be patterned after the 
following general statutory requirements: specific maximum time limits for re- 
plying to a grievance or claim filed by a party; specific maximum time limits 
in which the relief requested would either have to be granted or denied; priority 
processing of claims of an emergency nature, including those in which delay 
would endanger or create substantial risk of injury or damage to a person or 
group of persons; safeguards to ensure independent decision making within the 
state administrative process; de novo review in every case of state administrative 
determinations as to questions of fact and law; and, procedures for periodic 


review of the continued adequacy of state administrative procedures, as well as 
complaints about the integrity or fairness of a state's procedures. 

I. Introduction 

The Subject 

Immunity of the sovereign from liability for money damages in civil actions 
is among the most deeply rooted doctrines of our legal order. Mitigating the 
harsh consequences of denying compensation to victims of official misconduct 
under the sovereign immunity doctrine, a body of law has arisen permitting 
damage actions against officials where such misconduct is alleged. In addition, 
the sovereign immunity defense generally does not apply to municipalities and 
other units of local government charged with violations of federal law. Even in 
those instances where sovereign immunity has been waived or does not apply, 
liability for official misconduct may extend to not only the government, but also 
to the errant official. 

The existing system of civil sanctions for official misconduct thus relies 
heavily on the recovery of damages from officials to accomplish the goal of 
compensating the victims of such misconduct. The possibility of such recovery, 
in turn, is largely relied upon to deter further official misconduct and to assure 
the wrongdoer's accountability to the victim and to society at large. 

Notwithstanding the apparent logic of the existing system, most observers 
today agree that the system fails to serve two of its primary goals — measured 
deterrence of offical misconduct, and adequate compensation of the victims of 
such misconduct. On the one hand, the ever-present threat of damage suits is 
widely believed to deter not only improper conduct by officials, but proper 
conduct as well. On the other hand, there is general recognition that private 
damage actions against individual officials do not afford the victims of official 
misconduct financially responsible defendants. A further concern, at the local 
level, is the potentially crippling effect of large civil damage awards against 
municipalities in a period of recession and retrenchment. 

In recent years, the impetus for change of the civil sanction system has 
grown. This trend is fueled by the modern expansion of governmental activity 
that can give rise to allegations of harm, by the explosion of judicially recognized 
causes of action that can give rise to official liability as well as by judicial 
constriction of the umbrella of official immunity. Literally thousands of lawsuits 
are pending in which the issue of official immunity is critical. For ten years 
Congress has considered legislation that would revamp the existing law governing 
the liability of federal officials for actions taken in the course of their official 
duties. ' At the state and local levels of government there are tremendously varied 
responses to increasingly acute problems of official liability.^ 

1. See infra Chapter IV and legislative material compiled in Appendix A. 

2. See infra Chapter V and State Survey compiled in Appendix B. 


Inevitably, however, the debate over legislative alternatives to the existing 
system has been bedeviled by seemingly incompatible political considerations. 
Many government officials contend that the real problem to be addressed is not 
how to deter official misconduct or to compensate its victims, but rather how 
to free civil servants from the debilitating threat of civil damage actions, and 
municipal governments from the crushing burden of civil liability. Civil liberties 
groups and others, by contrast, focus on the importance of deterrence and ac- 
countability, and on the need to assure adequate compensation for the victims 
of government misconduct. Reconciliation of these competing concerns has thus 
far proved a most formidable challenge. 

Adding to the complications created by political disagreements among those 
favoring reform is the vastness and intricacy of the subject. The burgeoning 
literature devoted to the topic of sovereign immunity and official liability attests 
to the intense concern for the subject among scholars.^ Recent Supreme Court 
cases alone involve the immunity of executive branch officials ranging from the 
President"^ and his closest advisors^ to middle- and lower-level career civil ser- 
vants,^ members of the armed forces,^ legislative officials including members 
of Congress,*^ and quasi-judicial officers like federal prosecutors^ and adminis- 
trative law judges,'^ and finally, state and local officials in all branches of 

Moreover, fundamental questions about the ideal scope of government in- 
volvement in everyday life, and about the proper allocation of power between 
the federal government and the states, bear directly on proposed alternatives to 
the existing system of sovereign immunity and official liability. Indeed, it has 
been noted that the problems this study will address cannot be readily resolved: 
"The answers turn in part upon empirical data that are unavailable. They concern 
fundamental values about which there is little social agreement."'^ 

The AC US Mandate 

The Administrative Conference of the United States (ACUS) has initiated 
and sponsored this project in order to obtain comprehensive analysis and advice 

3. See infra Annotated Survey of the Literature compiled in Appendix C. 

4. Nixon V. Fitzgerald. 102 S. Ct. 2690 (1981). 

5. Harlow v. Fitzgerald, 102 S. Ct. 2727 (1982); Halperin v. Kissinger, 606 F.2d 1 192 (D.C. 
Cir. 1979), ajfd by an equally divided court, 101 S. Ct. 3132 (1981). 

6. Velde v. National Black Police Ass'n.. Inc., 631 F.2d 784 (D.C. Cir. 1980), remanded 
in light o/ Harlow, supra note 5 102 S. Ct. 3503 (1982). 

7. United States v. Brown, 348 U.S. 1 10 (1954); see also Coffey v. United States, 324 F. 
Supp. 1087 (S.D. Cal. 1971). 

8. Hutchinson v. Proxmire. 443 U.S. 1 1 1 (1979). See also Benford v. ABC, Inc., 502 F. 
Supp. 1148 (D. Md. 1980). 

9. Imbler v. Pachtman, 424 U.S. 409 (1976). 

10. Butz V. Economou, 438 U.S. 478 (1978). 

1 1 . E.g. Owen v. City of Independence, 445 U.S. 622 (1980); Maine v. Thiboutot, 448 U.S. 
1 (1980); Maher v. Gagne, 448 U.S. 122 (1980). 

12. Schuck, Suing Our Servants: The Court. Congress, and the Liability of Public Officials 
for Damages, 1980 S. Ct. Rev. 281, 282. 


on the law of official liability and immunity.'^ In executing the ACUS mandate, 
this project has been focused in a number of significant ways. First, consistent 
with ACUS' mandate and for practical reasons, it is worthwhile for this project 
to concentrate on the subject of executive rather than judicial or legislative 
immunity. The law of judicial immunity, although provocative, is relatively 
straightforward.'"^ Legislative immunity similarly is a less fertile area for this 

13. The research contract with ACUS describes the project as follows: 

The Contractor shall conduct a study, prepare a report, and shall draft proposed rec- 
ommendations on the subject of government officials' personal liability for violations of 
the Constitution and civil rights laws arising out of actions taken in the course of official 

The contract describes the scope of the work under the project as follows: 

The Contractor, in the course of preparing the report and recommendations shall: 

( 1 ) Briefly summarize the background of the problem, relevant case law, legal analyses 
and other articles. 

(2) Analyze the relative costs and benefits of the existing liability system. 

(3) Evaluate relevant proposals for legislation, including, without limitation, bills that 
would substitute the U.S. for the official, immunize the official, create administrative 
disciplinary systems, expand the Federal Tort Claims Act, and modify 42 U.S.C. § 1983. 

(4) Examine special problems facing particular groups of officials, including state and 
local employees or others involved with federal grant or revenue sharing programs, law 
enforcement officials, and high-level supervisory personnel, and recommend solutions. 

(5) Examine significant procedural questions that are likely to arise under the current 
system including, without limitation, choice of state or federal forum, governmental reim- 
bursement of officials' judgments and attorneys' fees, rules governing punitive damage 
awards, utility of 'preemptive' declaratory judgment suits, pleading and burden of proof 
issues, and recommend useful approaches to resolving them. 

The Contractor shall also participate in such meetings of the Conference, its committees, 
its Council, and the Chairman's staff as may be required by the Office of the Chairman, 
for consideration of the study and report that are the subject of this contract and resulting 
proposals, including plenary sessions and other meetings taking place after submission of 
and payment for the final report as provided herein. The Contractor shall cooperate with 
the staff of the Office of the Chairman as necessary and appropriate for the completion of 
this contract. 

14. Note the outrageous judicial conduct that was immunized by Stump v. Sparkman, 435 
U.S. 349 (1978). The Court reversed the seventh circuit which had denied immunity for a judge 
who approved (on the day it was filed) a mother's petition to have her daughter sterilized. (There 
was no arguable statutory basis for entertaining the sterilization petition, the hearing was ex parte, 
the only grounds for the order was the mother's uncorroborated statement that her 15-year old 
daughter was "somewhat retarded" and associated with young men, and there was no notice or 
possibility of appeal afforded to the girl (the girl, told she needed an appendectomy, was operated 
upon shortly after approval of the petition).) See also, Pierson v. Ray, 386 U.S. 547 (1967). 

As Justice Rehnquist sarcastically noted about judicial immunity: 

If one were to hazard an informed guess as to why such a distinction in treatment between 
judges and prosecutors, on the one hand, and other public officials on the other, obtains, 
mine would be that those who decide the common law know through personal experience 
the sort of pressures that might exist for such decision makers in the absence of absolute 
immunity, but may not know or may have forgotten that similar pressures exist in the case 
of nonjudicial public officials to whom difficult decisions are committed. But the cynical 
among us might not unreasonably feel that this is simply another unfortunate example of 


project to explore. Legislative immunity for Congressmen derives, of course, 
from the Speech or Debate Clause of Article I of the United States Constitution'^ 
and case law affords comparable immunity for state and local legislators.'^ 

Another refinement to this project that is appropriate in light of the ACUS 
mandate is to consider the ramifications of official immunity at the state and 
local level, but only insofar as the potential liability flows from federal assistance 
programs. Even this, in itself, is an appropriate subject for a separate research 

It is probably impossible to formulate rules sufficiently general for use in 
appraising existing or proposed substitute mechanisms for the control and de- 
terrence of official misconduct in the myriad of circumstances in which it may 
occur. Account must be taken of the unique issues raised in the many disparate 
contexts involved — from that of the cop on the beat, to that of the state social 
worker, the federal program manager, and the President of the United States. 
Any search for a simple solution to the problem in all of its contexts is sure to 
be disappointed. 

With these limitations in mind, the Authors hope that this study will avoid 
undue abstraction while attending to those complexities of the subject that have 
consistently eluded recognition. This project aims not only to suggest areas in 
which future study would be worthwhile, but to offer recommendations for such 
change as may clearly be warranted today. The starting point will be to describe 
the analytic framework of the study. 

judges treating those who are not part of the judicial machinery as lesser breeds without 
the law.' 

Butz V. Economou.438 U.S. 478. 528 n."*" (1978) (Rehnquist. J., concurring in part and dissenting 
in part, joined by Burger, C. J., Stewart, J., and Stevens, J.). But see Supreme Court of Virginia 
V. Consumers Union. 446 U.S. 719 ( 1980); Lopez v. Vanderwater, 620 F.2d 1229 (7th Cir. 1980). 

15. Hutchinson v. Proxmire, 443 U.S. 1 1 1 (1979); United States v. Helstoski, 442 U.S. 477 
(1979); Eastland v. United States Servicemen's Fund, 421 U.S. 491 (1975); Doe v. McMillan, 412 
U.S. 306 (1973); Gravel v. United States, 408 U.S. 606 (1972); United States v. Brewster, 408 
U.S. 501 (1972); Powell v. McCormack, 395 U.S. 486 (1969); Dombrowski v. Eastland. 387 U.S. 
82 (1967); United States v. Johnson, 383 U.S. 169 (1966); Tenney v. Brandhove. 341 U.S. 367 
(1951); Kilboum v. Thompson, 103 U.S. 168 (1881). 

In recent years, the Court has restricted the availability of Speech or Debate Clause immunity. 
The Court has refused to find legislators immune unless the challenged activity can be shown to 
have been not merely related to legislative activities but essential to the deliberations of the legislature. 
See Hutchinson v. Proxmire, 443 U.S. Ill, 130-33 (1979); see also Davis v. Passman, 442 U.S. 
228 (1979); Gravel v. United States. 408 U.S. 606 (1972); Cass. Damage Suits Against Public 
Officers. 129 U. Pa. L. Rev. 1110. 1131 n.96 (1981). And one very recent decision is of great 
concern to Congressional staff members. Benford v. ABC, Inc.. supra note 8. 

16. Tenney v. Brandhove. 341 U.S. 367 (1951); Lake Country Estate, Inc. v. Tahoe Regional 
Planning Agency, 440 U.S. 391 (1979). 

1 7 . See generally. Madden & Harkins. New Block Grant Programs Face Period of Adjustment 
in the Courts, Nat. L.J.. Mar. 8. 1982. at 29; Groszyk & Madden. Managing Without Immunity: 
The Challenge for State and Local Government Officials in the 80' s. Pub. Ad. Rev. (March/ April 
1981); Peters. Municipal Liability After Owen v. City of Independence and Maine v. Thiboutot, 13 
Urb. Law. 407 (1981); Freilich, 1979-1980 Annual Review of Local Gov't Law: Municipal Liability 
and Other Uncertainties. 12 Urb. Law. 557 (1980). 


11. Analytic Framework 

An appreciation of the complexity of the subject may be gleaned from the 
works of Schuck' and Cass,^ among others.^ Their analysis has revealed the 
extent of the underbrush, and provides a useful point of departure for this study. 
An essential first step is to sketch the conceptual framework distilled, for the 
purposes of this study, from the case law and literature. 

Key Concepts 

The terms "official immunity" and "official liability" are used to refer to 
the personal immunity and liability of government officials'^ in civil damage 
actions brought against officials in their individual capacities for actions taken 
in the course of employment or, in some cases, under color of office or law. 
The terms, correctly applied, do not embrace lawsuits arising from conduct 
ordinarily unrelated to the official's performance of his governmental duties, or 
where no action under color of office or law is involved. An obvious example 
of such conduct is a car accident involving a vacationing government official 
that leads to a suit seeking damages for the official's negligent driving. Claims 
of this sort have nothing whatsoever to do with any government function or 
purported exercise of legal authority.^ The terms "official immunity" and "of- 
ficial liability" also do not apply to causes of action for injunctive relief, which 
are deemed to be ultimately directed toward the government rather than to the 
individual holding an official position. 

The term "governmental immunity" refers to the doctrine of sovereign 
immunity. In the context of this report, the term "governmental liability" applies 
to situations in which the government has waived sovereign immunity and is 
derivatively liable for the conduct of officials. Treating the government respon- 
sibility as analogous to that of the employer under the doctrine of respondeat 
superior is, of course, only one way to look at the matter. Equally valid is the 
notion that the government's responsibility is not derivative but ultimate.^ 

\. Schuck, Suing Our Servants: The Court, Congress, and the Liability of Public Officials 
for Damages, 1980 S. Ct. Rev. 28L Professor Schuck's book on this subject is forthcoming in 
1983. See also Federal Tort Claims Act: Hearing on S. 1775 Before the Agency Administration 
Subcommittee of the Senate Judiciary Committee (Pt. 2), 97th Cong., 2d Sess. (1982) (testimony 
of Peter H. Schuck). 

2. Cass, Damage Suits Against Public Officers, 129 U. Pa. L. Rev. 1 1 10 (1981). 

3. E.g., Civil Liability of Government Officials, 42 Law & Contemp. Probs. 1 (winter 1978), 
see especially 8 (Mashaw), 46 (Baxter), and 62 (Epstein). 

4. As used in this report, the term "government official" encompasses officials, employees, 
and all kinds of executive branch personnel. 

5. The question of when certain very high ranking officials such as the Secretary of State or 
the Attorney General are ever completely "off-duty" is an interesting but relatively minor issue for 
the purposes of this study. 

6. A key issue to be considered in determining the extent of governmental liability for in- 
dividual misconduct is whether such liability should arise only when the alleged misconduct has 
been committed by a government official acting within the scope of his office or employment, or 
whether such liability should attach to misconduct committed by any person allegedly acting under 
color of office or law as well. See Comment, Constitutional Tort Remedies: A Proposed Amendment 
to the Federal Tort Claims Act, 12 Conn. L. Rev. 492, 535-36 (1980). 


The term "governmental wrongs" is used to refer to harm inflicted by 
government officials within the scope of their office or employment, or by persons 
acting under color of office or law, that causes a legally cognizable injury. This 
term is also necessarily imprecise, for there are any number of possible classi- 
fications of governmental wrongs.'' When a distinction is intended between 
constitutional and other legal harms this will be noted. 

This project's focus on the executive official at the federal, state, and local 
level is also broad. At each level of government, there are many instances in 
which the distinction blurs between executive, legislative, and judicial officials. 
And, moreover, there are numerous types of executive officials. As will be 
shown, the differences among the various kinds of officials — elected officials, 
political appointees, civil service or career government employees, members of 
the armed forces, officials with direct public contact such as law enforcement 
officers, and social workers, and of course, former officials of all kinds — may 
be critical with respect to the immunity issue. ^ For example, it may be useful 
to determine whether the incidence of litigation falls disproportionately on any 
particular type of official. And, of course, understanding the function and char- 
acteristics of particular officials is crucial for analyzing how various components 
of an alternative system of civil sanctions will affect their behavior. 

Competing Objectives, Core Values 

An appraisal of alternative systems for dealing with governmental wrongs 
involves a balancing of societal interests and values. These interests may be 
viewed from the perspectives of the victim, the official, and the public. 

For the victim, there are the immediate objectives of compensation and 
retribution, and the longer-range goal of deterrence. Fundamental principles of 
justice demand assurances of compensation or redress adequate to make the 
victim whole for wrongs suffered as a result of official misconduct. The objective 
of retribution, referred to by some as accountability and by others as punishment, 
is recognized as an independent interest of the victim. Indeed, it is this objective 
which may dominate the decision to sue officials when there is little prospect 
of recovering damages. 

From the official's perspective, the matter essentially is one of fairness to 
himself and his colleagues. Most government officials are dedicated, hardwork- 
ing, and well-intentioned, often making sacrifices in pay and other benefits to 
serve the public. From this standpoint, it is thought to be unfair to punish or 
impose sanctions on an individual carrying out his responsibilities as a govem- 

7. See, e.g., Mashaw, supra note 3, at 10-14; Cass, supra note 2, at 1139-41. 

8. See Schuck, supra note 1, at 293-95. It is beyond the scope of this study to examine in 
depth such special cases as the military, see, e.g., H.R. 3799, 97th Cong., 1st Sess. (1981), and 
federal public defenders, see, e.g., H.R. 3060, 97th Cong., 1st Sess. (April 7, 1981). See Tort 
Claims: Hearings on H.R. 24, H.R. 3060, and H.R. 3799 Before the Subcommittee on Administrative 
Law and Governmental Relations of the House Committee on the Judiciary, 97th Cong., 1st Sess. 
(1981). See Comment, Constitutional Tort Remedies: A Proposed Amendment to the Federal Tort 
Claims Act, supra, note 6, at 51 1-25. 


ment official, or to subject that individual to the threat of personal damage 
actions, simply by virtue of holding an official position and attempting to carry 
out governmental responsibilities. 

The public interest demands that an official who violates legal standards 
be held to account. For, as Justice Brandeis observed long ago: 

In a government of laws, existence of the government will be imperiled if 
it fails to observe the law scrupulously. Our government is the omnipresent 
teacher. For good or for ill, it teaches the whole people by its example.^ 

The very legitimacy of our system is ill-served when government officials 
are not held accountable for their misconduct. Indeed, in a nation deeply sus- 
picious of government power and imbued with the conviction that no person is 
above the law, the social interest of discerning and penalizing abuses of gov- 
ernmental power is a transcendent one.'° 

Legitimacy also involves the question of how and who is to set immunity 
rules and to administer them. For example, it may not be deemed legitimate to 
involve the judiciary to decide cases where they are required to second-guess 
executive policy decisions. And officials, however agreeable to the concept of 
accountability, may nevertheless deeply resent citizen review of official actions. ' ' 

Fairness is also essential to the public interest in coping with government 
wrongs. An individual official should be able to discern whether or not his 
activity is within standards that will be deemed to be acceptable. In the absence 
of clear guidelines, or if behavioral guidelines are routinely changed and applied 
retroactively, it would be considered unfair to the official to impose sanctions 
for, without more, failure to anticipate the applicable standard of conduct — even 
if, at the same time, it seemed fair to afford the injured party redress. 

Still another aspect of fairness from a systemic standpoint is whether there 
is a risk of sanctions being imposed on the official erroneously — either when 
the official has not actually violated a standard of conduct, or when the sanction 
is in some sense in excess of what would be deemed just under the circumstances. 

The public interest is also served by the overriding objective of achieving 
effective and responsible government. Thus, official liability systems ought to 
deter wrongdoing by officials, without discouraging officials from executing 
their duties in a decisive, selfless, and socially beneficial manner. The possibility 
that liability for one sort of official error could induce other errors in the opposite 
direction — a phenomena sometimes referred to as "overdeterrence" — has re- 
ceived exhaustive attention from the courts in particular.'^ Concern about the 

9. Olmstead v. United States, 277 U.S. 438, 485 (1928) (dissenting opinion). 

10. Mashaw, supra note 3. See also Vaughn, The Personal Accountability of Public Em- 
ployees, 25 Am. U.L. Rev. 85 (1975). 

11. See, e.g., Schuck, supra note 1, at 363-64 (discussion of police reaction to civilian 
review boards). 

12. See Cass, supra note 2, at 1153. See Spalding v. Vilas, 161 U.S. 483, 498 (1896); 
Gregoire v. Biddle, 177 F.2d 579, 581 (2d Cir. 1949), cert, denied, 339 U.S. 949 (1950). 


problem of overdeterrence has, in fact, been the mainstay rationale for the 
absolute immunity of certain officials.'^ 

There are numerous alternative formulations of the relevant objectives to 
be found in the literature and the case law. However, the general statement 
provided above will serve for the purposes of this study to provide a touchstone 
for appraisal of alternative systems of coping with governmental wrongs. That 
is, the existing civil sanction system and feasible alternatives will be evaluated 
in terms of whether they serve these objectives and values. 

There is no formal calculus for this task. It is well recognized that the 
interests and values related to official immunity often collide with one another. 
And the necessary tradeoffs may well involve comparing "apples and oranges." 
Any estimation of the preferred method of the optimal civil sanction system 
inescapably involves difficult judgments. 

Alternative Methods of Dealing With Governmental Wrongs 

This study evaluates four categories of methods of dealing with govern- 
mental wrongs. The present civil sanctions system relies to varying extent on a 
combination of all four types of responses. 

First, of course, is the private action seeking damages from individuals for 
violations of constitutional or other rights. This is the approach defined earlier 
as "official liability" and is presently the principal means for responding to 
governmental wrongs. 

Another approach for affording redress for such injuries is the private action 
seeking damages from the government itself. Viewed as a means by which society 
as a whole bears the cost of the transgressions of its public servants, this approach 
.can be referred to as "enterprise liability," a term borrowed from tort law, where 
it is traditionally related to private enterprise, but is now generally applied to 
government enterprises as well. Under such a system, liability is imposed upon 
the entity best suited, by reason of size, insurability and control, to minimize 
the cost of official wrongs.'"* Although at all levels of government there have 
been partial waivers of sovereign immunity, enterprise liability is not available 
for the redress of a large class of governmental wrongs such as constitutional 

The form of enterprise liability on which this report will focus is one in 
which the government is required to make direct payment once the fact of 
wrongdoing has been established. Government responsibility for such costs of 
litigation as attorneys' fees and damages are components of enterprise liability. 

13. Nixon V. Fitzgerald, 102 S. Ct. 2690. 2703 (1982) (the President): Stump v. Sparkman, 
435 U.S. 349. 356-64 (1978) (judges); Tenney v. Brandhove, 341 U.S. 367, 377-78 (1951) 

14. See, e.g.. Schuck, supra note 1. at 288. 



Alternative forms of enterprise liability that will be discussed are insurance'^ 
and indemnity'^ systems. 

For the purpose of compensating victims, enterprise liability has an advan- 
tage over official liability: the governmental "enterprise" is much more likely 
than the individual officer to have the funds required to satisfy a judgment.'^ 
This, however, could not be the only advantage of a system of enterprise liability. 
Enterprise liability can tend to decrease the overdeterrence or "chill" placed on 
the vigorous decisionmaking process of a government official who is afraid to 
act because of the overhanging threat of personal damage actions. 

Two problems arise, however, when the governmental "enterprise" is re- 
quired to satisfy judgments. First, the potential for w/2(i^rdeterrence exists where 
the threat of personal liability no longer confronts governmental officials. Whereas 
the threat of individual liability induce officials to modify their behavior, en- 
terprise liability, by contrast, only indirectly encourages officials to modify their 
behavior; largely to the extent that other officials, through a system of admin- 
istrative controls, choose to require.'^ 

A third important category of responses to governmental wrongs may be 
referred to as internal administrative controls. This category includes mechanisms 
of control within the executive branch; i.e., a wide range of inducements to 
good behavior and sanctions for bad behavior including job termination or trans- 
fer, promotion or demotion, and changes in pay and perquisites. 

Needless to say, the array of internal incentives and sanctions comprising 
the administrative system of control, like the other alternatives, serves some 
goals more effectively and others less effectively. For example, although dis- 
ciplinary proceedings focus on the acts of a particular official, and may serve a 
deterrent or retributive function, they do not, standing alone, afford compensation 
to the victims of governmental wrongdoing. Moreover, these devices may also 
make it possible for agency heads to shift responsibility for misconduct from 
their own policies to the poor judgment or wrongdoing of an individual selected 
to serve as a scapegoat. 

A most significant characteristic of existing administrative methods of con- 
trolling governmental wrongdoing is that, unlike private damage actions against 
officials or the government, disciplinary sanctions generally cannot be instituted 
by the victim. In the current system, injured individuals must rely on others to 
initiate such actions. Though they may complain, petition, or use publicity to 
urge that disciplinary action be taken, they cannot themselves compel the ini- 
tiation of internal administrative control procedures, or act as parties thereto. 

15. Insurance is a form of loss shifting in which the government shares the risk of loss with 
a third party, the insurer. 

16. Indemnification involves government liability in private damage actions where the indi- 
vidual official is a nominal defendant. 

17. Cass, supra note 2, at 1174. 

18. Id. at 1175. 


The final type of response amounts to a variety of external controls that 
originate outside the executive branch. These controls include not only judicial 
review of official action but also legislative oversight — up to and including the 
remedy of impeachment — as well as the expression of public sentiment by means 
of the ballot. In contrast to administrative controls, external controls are less 
certain and clearly defined, and often depend heavily upon the level of publicity 
regarding official misconduct. Although, by themselves, legislative, judicial, or 
electoral controls are unlikely to prove an effective response to official miscon- 
duct, the symbiotic interplay of such external controls with the other instruments 
of the civil sanction system must be considered in appraising needed modifi- 

Having painted the foregoing schematic in broad strokes, a portrait of the 
existing civil sanction system may be clearer. 

official liability 219 

111. The Existing System of Dealing with Governmental Wrongs 

An understanding of the present civil sanction system is essential. The first 
section below describes the private suit for damages in terms of both liability 
and immunity of officials. The following sections describe the extent that the 
current civil sanction system involves alternative methods of responding to gov- 
ernmental wrongs: i.e., enterprise liability, administrative disciplinary controls, 
and finally, external controls. 

A. The Private Damage Action 

1. Expansion of Individual Liability For Money Damages 

The last decade has witnessed a significant expansion in the vulnerability 
of executive branch officials at the federal, state and local levels to personal 
liability for money damages. A number of factors have led to the greatly increased 
number of actions brought against government officials during this time. A most 
significant factor that hardly requires elaboration is the extraordinary growth of 
government. Simply put, expanded government operation coincides with a higher 
incidence of harmful official action of all kinds — both real and perceived — that 
generates suits alleging a variety of legal wrongs. Here it will be assumed that 
the scope of government is a given variable. 

Other contributing factors merit consideration. In general, the well-recog- 
nized trend towards an expansion in judicially-discovered rights results in in- 
creased claims alleging violation of constitutional rights. The broadening of the 
section 1983 action, as well as the recognition and expansion of the direct Bivens 
action also play major roles in this expansion of zones of potential individual 
liability for money damages. In addition, the recently increased ability to recoup 
court costs and attorneys' fees in section 1983 actions has provided an incentive 
for plaintiffs to bring actions against executive officials at the state and local 
level. Finally, the broadened availability of information about government ac- 
tivities through channels such as the Freedom of Information Act has made it 
easier to establish the factual predicate for litigation against officials. Some of 
the most significant influences on the evolving nature of the private damage 
action against officials are briefly reviewed below. 

— Judicially Discovered Rights — General Trends 

During the early 1970's, the Supreme Court greatly widened the procedural 
due process rights designed to protect the individual against government in- 
fringement.^ The Court created new "property" and "liberty" rights in rec- 
ognition of citizens' growing reliance on government for benefits and other 
entitlements.^ In doing so, the old distinction between privileges and rights was 

1. For a discussion ofthisjudicial expansion of recognized rights, 5ee, e.g., L. Tribe, American 
Constitutional Law § 10-9 (1978). 

2. Charles Reich's article, The New Property, 73 Yale L.J. 733 (1964) played a significant 
role in recognizing the increased dependence of many citizens upon the state. 


rejected. As Professor Tribe has written, "[f]or the first time, the Court rec- 
ognized as entitlements interests founded neither on constitutional nor on com- 
mon-law claims of right but on a state-fostered (and hence justifiable) expectation."^ 
For example, the Court recognized an individual's property right to welfare 
benefits once welfare assistance has been initiated (thus the benefits cannot be 
terminated without a hearing),"^ and a prisoner's liberty interest in parole once 
an early release is promised.^ The Court also recognized a property interest in 
government employment where continued employment is promised expressly or 
through implication.^ 

Furthermore, the early 1970's saw an expansion in what some describe as 
the "core" procedural due process interests, as well as the creation of the new 
entitlements. Thus, under some circumstances, the government must meet due 
process requirements before maligning a person's good name or reputation.^ The 
Court also built a type of privacy interest into the core property interest by 
enabling people to be secure against seizure of their personal belongings by the 
state or another party with state approval.^ In the latter half of the 1970's, the 
Court reversed this liberalizing trend and began to narrow the individual's liberty 
and property rights protected by procedural due process.^ These rights, however, 
certainly remain far greater than they did a decade ago. 

—Section 1983 Actions^^ 

In addition to the increase in judicially discovered rights, the courts in recent 
years have eased the requirements for plaintiffs bringing damage suits based on 
those rights. In the last two decades, the number of actions brought against state 
and local government officials under 42 U.S.C. § 1983 has increased nearly one 
hundred-fold. ' ' That provision, enacted in the Civil Rights Act of 187 1 , provides 
for liability in a civil action for any person who under color of law deprives 
another person of "any rights, privileges or immunities secured by the Consti- 
tution and laws." 

Several recent decisions have increased the vulnerability of state and local 
government officials to suits brought under section 1983. For example, in Maine 
V. Thiboutot,^^ the Supreme Court held that the phrase "and laws" in that 

3. Tribe, supra note 1, at 515. 

4. Goldberg v. Kelly, 397 U.S. 254 (1970). 

5. Morrissey v. Brewer, 408 U.S. 471 (1972). The same entitlement interest in probation 
was recognized in Gagnon v. Scarpelli, 411 U.S. 778 (1973). 

6. See Perry v. Sindermann, 408 U.S. 593 (1972); Board of Regents of State Colleges v. 
Roth, 408 U.S. 564(1972). 

7. Wisconsin v. Constantineau, 400 U.S. 433 (1971). 

8. Tribe, supra note 1, at 520. 

9. See id., §§ 10-11. 

10. An expanded analysis of the section 1983 case law, with an emphasis on liability under 
federal asssistance, is contained in Chapter V. 

11. In 1979, there were neariy 25,000 filings in federal courts, as compared to 261 in I96I. 
Groszyk & Madden, Managing Without Immunity: The Challenge for State and Local Government 
Officials in the 1980' s, Pub. Admin. Rev. (March/ April 1981). 

12. 448 U.S. 1 (1980). 


provision refers to all federal statutes, and not just those pertaining to civil rights. 
This broad interpretation will undoubtedly increase the number of section 1983 
suits filed against state and local government officials for violations of federally 
guaranteed rights.'-^ 

Other recent decisions open the door to more suits against government 
officials. In Parratt v. Taylor, ^^ decided in 1981 , the Court decided that section 
1983 is not limited to intentional deprivations of constitutional rights. One year 
later, the Court held that a plaintiff is not required to exhaust state administrative 
remedies prior to the filing of a section 1983 action. ^^ 

The trend toward increased section 1983 litigation could be slowed if, in 
the wake of Harlow v. Fitzgerald,^^ discussed below, it becomes easier for state 
and local officials to establish a good-faith defense. But this result is hardly 
inevitable. And notwithstanding Harlow, the defendant official would probably 
maintain the burden of pleading his good-faith defense.'^ In most circuits the 
defendant official also bears the burden of persuasion.'^ 

— Bivens Actions 

In addition to the expansion of liability for money damages for state and 
local officials under section 1983, the number and types of actions against federal 
officials has also greatly increased in the decade since the Supreme Court's 1971 
decision in Bivens v. Six Unknown Named Agents. ^"^ In Bivens, the Supreme 
Court held that a plaintiff may bring an action for money damages against a 
federal official who violates his Fourth Amendment rights. Despite the lack of 
any statutory cause of action, the court expressly found in Bivens that a cause 
of action may be premised upon a violation of the Constitution. 

The number of direct ''Bivens actions" brought has increased dramatically, 
with law enforcement and prison officials as the most frequent defendants. -^^ 

13. In his dissent. Justice Powell stated "[n]o one can predict the extent to which litigation 
arising from today's decision will harass state and local officials; nor can one foresee the number 
of new filings in our already overburdened courts. But no one can doubt that these consequences 
will be substantial." Id. at 23. Federal statutes will not automatically allow a plaintiff to sue under 
section 1983, however — the courts now must determine whether the statute secures rights which 
can be litigated in section 1983 claims. The Supreme Court has recently held that where "the 
remedial devices in a particular act are sufficiently comprehensive, they may suffice to demonstrate 
a congressional intent to preclude the remedy of suits under § 1983." Middlesex County Sewage 
Authority v. National Sea Clammers Associations, 453 U.S. 1, 20 (1981). 

14. 101 S. Ct. 1908 (1981). 

15. Patsy v. Board of Regents, 102 S. Ct. 2557 (1982). 

16. 102 S. Ct. 2727 (1982). 

17. See Gomez v. Toledo, 446 U.S. 635 (1980). 

18. See Cass, Official Liability: Damage Suits Against Public Officers, 129 U. Pa. L. Rev. 
1110, 1129 n.90 (1981) (cases collected). 

19. 403 U.S. 388(1971). 

20. See, e.g., Federal Tort Claims Act: Hearings on S. 1775 Before the Subcommittee on 
Agency Administration of the Senate Committee on the Judiciary (Pt. 1), 97th Cong., 1st Sess. 4 
(1981) (testimony of Deputy Attorney General Edward C. Schmults). Cass reports nearly a fivefold 
increase in the legal actions filed by prisoners alleging violations of constitutional guarantee in the 
decade from 1970 to 1980. Cass, supra note 18, at 1159. (Based on figures of the Administrative 
Office of the United States Courts). 


One reason for the proliferation of direct actions for violations of consti- 
tutional rights has been the courts' expansion of the so-called Bivens action. 
Although the Supreme Court restricted its holding in Bivens to the fourth amend- 
ment's guarantee against unreasonable searches and seizures, some lower federal 
courts have since extended the theory to the first, fifth, sixth, eighth, ninth, 
tenth, thirteenth and fourteenth amendments.-^' In Davis v. Passman, ^^ the Su- 
preme Court held that a Bivens action may be brought directly on the due process 
clause of the fifth amendment. The Court, citing Bivens, reasoned that a damages 
remedy was appropriate and that there was no congressional declaration that 
money damages should not be awarded. ^^ 

In addition, the Supreme Court held in Carlson v. Green that a Bivens 
remedy is available even where an alternative remedy against the federal gov- 
ernment under the Federal Tort Claims Act (FTC A) is available. ^"^ The Court 
reasoned that four advantages of the Bivens action indicated that Congress did 
not intend to limit a plaintiff to the FTC A remedy. First, it concluded that since 
the Bivens remedy is recoverable against individuals, it serves as a more effective 
deterrent than the FTC A remedy against the federal government. Second, pu- 
nitive damages may be awarded in a Bivens suit but are statutorily prohibited in 
a FTC A suit. Third, a plaintiff cannot opt for a jury trial in a FTC A action. 
Finally, an action under the FTC A exists only if the state in which the alleged 
misconduct occurred would permit a cause of action for that misconduct to go 
forward. ^^ 

Some courts have also applied Bivens in suits alleging unconstitutional 
conduct by state officials, despite the availability of alternative remedies such 
as section 1983, and other courts, going even further, have allowed fi/ve«5 actions 
against private individuals or entities. ^^ 

— Litigation Costs — Attorneys' Fees 

It is widely recognized that the willingness to litigate claims against gov- 
ernment officials is also related to the amount and incidence of the cost of 

21. See, e.g., Lehmann, Bivens and its Progeny: the Scope of a Constitutional Cause of 
Action for Torts Committed by Government Officials, 4 Hastings Const. L.Q. 531, 566-72 (1977); 
note, Bivens and the Creation of a Cause of Action for Money Damages Arising Directly From the 
Due Process Clauses, 29 Emory L.J. 231 (1980). 

22. 442 U.S. 228 (1979). 

23. Id. at 245-47. 

24. 446 U.S. 14 (1980). The basic provisions of the Federal Tort Claims Act appear in 28 
U.S.C. § § 1346(b), 2671-2680. 

25. 446 U.S. at 20-23. Following Bivens, the FTCA was amended to waive sovereign 
immunity and allow suits against the federal government arising out of assault, battery, false arrest, 
false imprisonment, malicious prosecution, and abuse of process committed by law enforcement 
officers. 28 U.S.C. § 2680(h). See Boger, Gitenstein, & Vertvil, The Federal Tort Claims Act 
Intentional Torts Amendment: An Interpretive Analysis, 54 N.C.L. Rev. 497 (1976) (reprinting 
excerpts from legislative history). 

26. See Lehmann, supra note 21, at 575-87. 


litigation. -^^ The proliferation of section 1983 actions against state and local 
government officials is partially explained by the availability of attorneys' fees 
for successful plaintiffs. In 1976, Congress passed the Civil Rights Attorneys' 
Fees Awards Act of 1976, thereby authorizing courts to award attorneys' fees 
to the prevailing party in actions to enforce the terms of the Civil Rights Act of 
1866, which includes section 1983.^^ Moreover, in Maker v. Gagne}^ the 1980 
companion case to Thiboutot, the Supreme Court held that this attorneys' fees 
provision applies to all types of section 1983 actions, and not just to those based 
on a violation of the Constitution or a federal statute providing for the protection 
of civil rights. In that same case, the Court also decided that a plaintiff may be 
awarded attorneys' fees even if he prevails through a settlement rather than 
through litigation. ^^ 

Under the so-called "American Rule," such awards are not recoverable in 
Bivens actions. The FTC A does not provide for fees and costs. 

— Litigation Costs — Government' s Defense of Officials 

When a federal official is sued for actions taken within the scope of his 
official duty, he will ordinarily be represented by Justice Department attorneys, 
at government expense.^' This is also generally the case throughout the states. 
This representation of federal employees will usually be provided regardless of 
whether the suit is brought in state or federal court, as long as the official is not 
the target of a federal criminal investigation concerning those actions. The official 
must make his request for representation to his superior in the agency in which 
he works. The agency will then forward the request to the Justice Department, 
along with a statement of whether the employee was acting within the scope of 
his employment.''^ 

In addition to the exception for officials subject to a federal criminal in- 
vestigation, the Justice Department will not defend an official in two other 
situations: when a vigorous defense will conflict with other interests of the United 
States, or when the defenses of several defendant officials in the action are 

27. It seems desire for retribution or harrassment motivates suits independently of the likely 
amount of a damage award or the cost of litigation. It is assumed that litigants are no angrier now 
than they ever have been and thus that this factor does not explain the increased amount of damage 
actions against officials. 

28. Codified at 42 U.S.C. § 1988 (1976). Successful defendants may also be reimbursed, 
but the standards are much more stringent than for plaintiffs. See Cass, supra note 18, at 1155 n. 
70. See generally, Witt, The Civil Rights Attorneys' Fees Awards Act of 1976, 13 Urb. Law. 589 

29. 448 U.S. 122(1980). 

30. Id. See also Tatro v. Texas, 516 F. Supp. 968 (N.D. Tex. 1981). 

3 1 . Certain plaintiffs are able to obtain low cost legal aid or pro bono representation. However, 
officials have probably greater access to relatively free defense counsel in most cases. 

32. United States Department of Justice, Monograph. Damage Suits Against Federal Officials; 
Department of Justice Representations Immunity 1 1 (Revised, November 1981). See also 28 C.F.R. 
§§ 50.15, 50.16 (1981); United States Attorneys' Manual, Title 4, HH 4-13.000, 13.300 et seq. 


inconsistent. In such cases, the Justice Department will retain private counsel 
for the official at public expense.^'' 

For state and local officials sued in their individual capacity under section 
1983, state or local law will govern the availability of public counsel (or public 
payment) for the defense. A majority of states do provide for public defense if 
the actions upon which the suit is based were taken within the scope of the 
officials' employment. ^"^ 

2. The Immunity Doctrine 

— Evolution of the Case Law 

Tracing the development of the immunity doctrine reveals that the courts 
have used federal, state, and local precedents interchangeably, and even a cursory 
analysis of the evolution of the case law also demonstrates the difficulties en- 
countered by the judiciary in wrestling with the public policy issues in this field. 

The doctrine of official immunity, which some contend is an essential 
underpinning to our form of government, did not reach its zenith until the mid- 
twentieth century. "^^ Professor Mashaw explains: 

Early opinions made no distinction between public officers and ordinary 
citizens in determining liability for tortious conduct. Indeed, many early 
decisions imposed liability for official acts that were not even tortious by 
private-law standards [E.g., Miller v. Norton, 152 Mass. 540, 26 N.E. 100 
(1891) (Holmes, J.)]. 

. . . Nineteenth century decisions often afforded public officers little pro- 
tection against liability even for actions taken in the performance of their 
official duties. ^^ 

A departure from this early pattern was anticipated by the 1872 case of 
Bradley v. Eisher,^^ where the Supreme Court established absolute immunity 
forjudges acting in their judicial capacity, so long as there was subject matter 

33. See Bell. Proposed Amendments to the Federal Tort Claims Act, 16 Harv. J. on Legis. 
1, 8-9 (1979). Between 1976 and 1979, the Justice Department retained approximately 75 private 
firms for such representation, pursuant to 28 C.F.R. § 50. 16, at a cost in excess of $2 million. Id. 
Critics of the controversial private counsel program maintain that it is costly, extends the duration 
of cases, and makes them more difficult to settle. Id. 

34. See infra Chapter V and Appendix B. 

35. A provocative and interesting colloquy on the antecedents and symbiotic development of 
sovereign immunity and official immunity is reprinted at 42 Law & Contemp. Prob. 82-89 (Winter 

36. Mashaw, Civil Liability of Government Officials. 42 Law & Contemp. Prob. 1 Part II, 
at 14-15, n.3. In a quote from Engdahl. Immunity- and Accountability for Public Governmental 
Wrongs, 44 U. Colo. L. Rev. 1, 47 (1972), Mashaw notes "[t]hese standards of personal official 
liability were repeatedly reaffirmed and applied during the same decade around the turn of the century 
when the Supreme Court was enlarging the immunity of the state; indeed it was only for this reason 
that the expanding state immunity was considered to be consistent with the tradition of effective 
redress for positive governmental wrongs." 

37. 80 U.S. (13 Wall.) 335 (1872). 



jurisdiction.^^ Even malicious or corrupt motives would not subject a judge to 
liability. ^^ In providing this broad cloak of immunity, the Court stressed the 
importance of absolute judicial freedom: 

[I]t is a general principle of the highest importance to the proper admin- 
istration of justice that a judicial officer, in exercising the authority vested 
in him, shall be free to act upon his own convictions, without apprehension 
of personal consequence to himself. Liability to answer to everyone who 
might feel himself aggrieved by the action of the judge, would be incon- 
sistent with the possession of this freedom, and would destroy that inde- 
pendence without which no judiciary can be either respectable or useful."*^ 

Subsequently, high-level executive officers were also deemed entitled to 
absolute immunity from civil suit, where their alleged wrongful acts were com- 
mitted in the course of the discharge of their duties. In Spalding v. Vilas, ^^ the 
Supreme Court reasoned that the public policy considerations which exempt 
judges from civil liability also apply to the heads of executive departments. As 
in Bradley, the Court recognized ''a distinction between action taken by the head 
of a department in reference to matters which are manifestly or palpably beyond 
his authority, and action having more or less connection with the general matters 
committed by law to his control or supervision.'"^^ 

This broad protection from liability received an additional extension from 
the often quoted decision by Judge Learned Hand in Gregoire v. Biddle,^^ decided 
over 50 years after Spalding. Not only did the Court hold that the officials' 
alleged maliciousness would not, if proven, subject them to liability, but Judge 
Hand also apparently eliminated the distinction for officials acting without legal 
authority. '^'^ Furthermore, Biddle broadened the doctrine of absolute immunity 
for heads of executive departments to cover mid-level executive officials (in 

38. The Court distinguished cases in which there clearly was no jurisdiction. In the latter 
case, "any authority exercised is usurped authority, and for the exercise of such authority, when 
the want of jurisdiction is known to the judge, no excuse is permissible." Id. at 352. Thus, the 
judge would be subject to civil liability. Cf. Stump v. Sparkman, 435 U.S. 349 (1978). 

39. 80 U.S. (13 Wall.) at 352. 

40. Id. at 347. This absolute protection for the judicial officers from liability in civil suits 
remains today, despite the evolution of only a qualified immunity for executive officers. Critics 
have maintained that no legitimate reason exists for this distinction, except perhaps the willingness 
on the part of judges to be more protective of themselves than of others who bear similar respon- 
sibilities and pressures. E.g., Butz v. Economou, 438 U.S. 478, 528 n. "*" (1978), quoted supra 
Chapter I, note 14. At this time, absolute immunity is also available to judicial officers like pros- 
ecutors, legislators, and the President of the United States. See Nixon v. Fitzgerald, 102 S. Ct. 2690 

41. 161 U.S. 483 (1896). 

42. Id. at 498. 

43. 177 F.2d 579 (2d Cir. 1949). 

44. Hand asserted, "[w]hat is meant by saying that the officer must be acting within his 
power cannot be more than that the occasion must be such as would have justified the act, if he 
had been using his power for any of the purposes on whose account it was vested in him." Id. at 


Biddle, the Director of the Enemy Alien Control Unit of the Justice Department 
and a district director of Immigration were accused of false arrest and impris- 

The Supreme Court explicitly extended this protection from civil liability to 
lower-level executive officers in Barr v. Matteo^^ The Court stated: 

We do not think that the principle announced in Vilas can properly be 
restricted to executive officers of cabinet rank, and in fact it never has been 
so restricted by the lower federal courts. The privilege is not a badge or 
emolument of exalted office, but an expression of a policy designed to aid 
in the effective functioning of government. The complexities and magnitude 
of governmental activity have become so great that there must of necessity 
be a delegation and redelegation of authority as to many functions, and we 
cannot say that these functions become less important simply because they 
are exercised by officers of lower rank in the executive hierarchy. 

To be sure, the occasions upon which the acts of the head of an 
executive department will be protected by the privilege are doubtless far 
broader than in the case of an officer with less sweeping functions. But that 
is because the higher the post, the broader the range of responsibilities and 
duties, and the wider the scope of discretion, it entails. It is not the title of 
his office but the duties with which the particular officer sought to be made 
to respond in damages is entrusted. . . ."^^ 

Thus, since the director of the Office of Rent Stabilization acted "within the 
outer perimeter of [his] line of duty'"^^ in allegedly libeling his employees, he 
was protected by absolute immunity. 

This immunity from liability was held to apply to local government officials. 
In its 1967 decision in Pier son v. Ray,"^^ the Supreme Court addressed the question 
of the liability of a judge and police officers who allegedly, under color of state 
law, infringed the plaintiff's federally guaranteed rights. The plaintiff joined his 
section 1983 suit with an action for false arrest and imprisonment. In addition 
to finding that the judge was immune from prosecution,"^^ the Court held that 
section 1983 did not remove the police officers' defense of good faith and 
probable cause which they could assert in the common-law action for false arrest 
and imprisonment. They could therefore claim the same immunity for the section 
1983 count as they could for the state law count. ^^ 

The enormous change in the area of official immunity during the past decade 
began in the 1974 landmark case of Scheuery. Rhodes. ^^ There, the estates of 
the students killed at the demonstration at Kent State University brought a section 
1983 action against the Governor of Ohio, the university president, and various 

45. 360 U.S. 564(1959). 

46. Id. at 572-73 (footnotes omitted) (Justice Harlan, writing for a plurality of four). 

47. Id. at 575. 

48. 386 U.S. 547(1967). 

49. The Court cited Bradley v. Fisher, supra note 37, in so holding. 

50. 386 U.S. at 555-57. 

51. 416 U.S. 232 (1974). 



officers and enlisted members of the Ohio National Guard. ^-^ The Court decided 
that only a limited or qualified good-faith immunity applies to both officers of 
the executive branch of government. In elaborating on this new standard, the 
Court stated: 

[I]n varying scope, a qualified immunity is available to officers of the 
executive branch of government, the variation being dependent upon the 
scope of discretion and responsibilities of the office and all the circumstances 
as they reasonably appeared at the time of the action on which liability is 
sought to be based. It is the existence of reasonable grounds for the belief 
formed at the time and in light of all the circumstances, coupled with good- 
faith belief, that affords a basis for qualified immunity of executive officers 
for acts performed in the course of official conduct. ^^ 

By affording the state officials the absolute immunity recognized in previous 
decisions, "§ 1983 would be drained of meaning," the Court reasoned. ^"^ 

In the eight years since Scheuer, the Court continued to whittle away the 
immunity from civil suits for executive officials. In Wood v. Strickland, ^^ the 
Court decided that the immunity standard contains elements of both an objective 
and a subjective good-faith test. Thus, a school official would not be protected 
simply on the basis of his good intentions. Instead, the standard would also be 
based on "knowledge of the basic, unquestioned constitutional rights of his 
charges. "^^ If a school official knew or reasonably should have known that his 
actions violated a student's rights, or if he acted maliciously, he would be subject 
to liability. The Court announced the same standard for superintendents of state 
hospitals in O'Connor v. Donaldson^^ and for prison officials in Procunier v. 

Despite the narrowing of the exemption for executive officials, however, 
prosecutors remain absolutely immune for actions taken in initiating and pursuing 
a criminal prosecution. This complete protection from liability for actions taken 
within the scope of prosecutorial duties has been described as a form of "quasi- 
judicial" immunity which was consistently recognized at common law.^^ In 
Imbler v. Pachtman,^ the Supreme Court held that this absolute immunity also 

52. The plaintiffs charged that these officials, acting under color of state law, "intentionally, 
recklessly, willfully and wantonly" caused an unprecedented National Guard deployment on the 
campus and ordered the Guard members to perform allegedly illegal acts resulting in the students' 
deaths. Id. at 235. 

53. Id. at 247-48. 

54. Id. at 248. 

55. 420 U.S. 320 (1975) (§ 1983 action brought by students against school officials, charging 
infringement of due process under color of state law for expulsion from school). 

56. Id. at 322. 

57. 422 U.S. 563 (1975). 

58. 434 U.S. 555 (1978). 

59. See, e.g., Tyler v. Witkowksi, 511 F.2d 449 (7th Cir. 1975); Kostal v. Stoner, 292 F.2d 
492 (10th Cir. 1961), cert, denied, 369 U.S. 868 (1962). 

60. 424 U.S. 419 (1976). Legislators and judges also retain their absolute immunity. Absolute 
immunity has been extended to legislative aides as well. Gravel v. United States, 408 U.S. 606 




exists in section 1983 suits. In addition, agency officials who perform functions 
analogous to those of a prosecutor are entitled to absolute immunity. ^^ 

During the past several terms, the Court has addressed the issues of what 
type of immunity should be granted to high federal executive officials, and to 
the President himself. Federal cabinet officials, including officers, are only to 
be awarded the Scheuer qualified good-faith immunity, the Court decided in 
Butz V. Economou.^^ 

In Economou, the plaintiff filed an action against a number of officials in 
the Department of Agriculture and other federal agencies, including the Secretary 
of Agriculture, alleging that they had instituted an investigation and administra- 
tive proceedings against him for his criticism of the Department. The defendants 
asserted that, as federal officials, they were entitled to absolute immunity for 
all discretionary acts within the scope of their authority. In rejecting this argument i 
and extending the Scheuer holding to cover federal officers. Justice White wrote ' 
for the majority: 

[I]n the absence of congressional direction to the contrary, there is no basis 
for according to federal officials a higher degree of immunity from liability 
when sued a constitutional infringement as authorized by Bivens than is 
accorded state officials when sued for the identical violation under § 1983. 
The constitutional injuries made actionable by § 1983 are of no greater 
magnitude than those for which federal officials may be responsible. The 
pressures and uncertainties facing decisionmakers in state government are 
little if at all different from those affecting federal officials. . . . Surely, j- 
federal officials should enjoy no greater zone of protection when they violate 
federal constitutional rules than do state officers [emphasis in the original] .^^ 

The Court in Economou also refused to recognize absolute immunity for 
cabinet officers, severely limiting its 1896 holding in Spalding v. Vilas.^ Justice j 
White, in addition to stating that our system of jurisprudence rests on the as- 
sumption that all individuals are subject to federal law, reasoned that "the greater 
power of such officials affords a greater potential for a regime of lawless con- 
duct. "^^ But the Court left open the possibility that absolute immunity might 
be granted to certain officials exercising discretionary authority, where such 
protection is essential for the proper functioning of government.^ ^ 

In the recent case of Nixon v. Fitzgerald, ^^ however, a sharply divided (5- 
4) Court held that the President is entitled to absolute immunity. Justice Powell, 

61. Butz V. Economou, 438 U.S. 478 (1978). 

62. Id. Four justices dissented on this point. 

63. M at 500-601. 

64. The dissent criticized the majority's "unnaturally constrained reading" of Spalding and 
stated that ""Spalding clearly and inescapably stands for the proposition that high-ranking executive 
officials acting within the outer limits of their authority are absolutely immune from suit." Id. at 
518, 519. 

65. Id. at 506. 

66. Id. at 507. 

67. 102 S. Ct. 2690 ( 1982). In Kissinger v. Halperin, 452 U.S. 713 (I98I), an equally divided 
Court failed to decide the issue. 



writing for the majority, considered Presidential immunity "a functionally man- 
dated incident of the President's unique office, rooted in the constitutional tra- 
dition of the separation of powers and supported by our history. "^^ This shield 
from civil liability extends for all acts within the "outer perimeter" of the 
President's official responsibility.^^ Despite a vigorous dissent, the Court rea- 
soned that alternative remedies for protection against Presidential misconduct 
are available, such as impeachment, scrutiny by the press, and oversight by 
Congress. ^° 

In Harlow^ v. Fitzgerald, ^^ the companion case to Nixon v. Fitzgerald, the 
Court refused to extend this absolute immunity to high-ranking White House 
officials. The Court rejected the officials' reliance on Gravel v. United States,^^ 
where the immunity for legislators was extended to cover their aides. Relying 
on Economou, the Court stated that such a derivative immunity for Presidential 
aides would sweep too wide and could not be reconciled with its "functional" 
approach to immunity law.^^ The Court did leave open the possibility that "[f]or 
aides entrusted with discretionary authority in such sensitive areas as national 
security or foreign policy, absolute immunity might well be justified to protect 
the unhesitating performance of functions vital to the national interest. "^"^ Gen- 
erally, however, only a qualified, good-faith immunity will apply. 

Most importantly, the subjective and objective analysis called for the test 
for this good- faith immunity was altered from that outlined in Wood v. Strickland. 
Concerned that judicial inquiry into subjective motivation may be disruptive of 
effective government, the Court decided that "bare allegations of malice should 
not suffice to subject government officials either to the costs of trial or to the 
burdens of broad-reaching discovery. "^^ Thus, the Court removed the subjective 
element from the good-faith defense. The Court stated: 

Reliance on the objective reasonableness of an official's conduct, as mea- 
sured by reference to clearly established law, should avoid excessive dis- 
ruption of government and permit the resolution of many insubstantial claims 
on summary judgment. On summary judgment, the judge appropriately may 
determine, not only the currently applicable law, but whether that law was 
clearly established at the time an action occurred. If the law at that time 

68. 102 S. Ct. at 2701. The Court added, "[b]ecause of the singular importance of the 
President's duties, diversion of his energies by concern with private lawsuits would raise unique 
risks to the effective functioning of government." Id. at 2703. 

69. Id. at 2705. The President's reorganization of the Air Force, in which plaintiff was dimissed 
from his job (he alleged it was in retaliation for testimony before Congress on military cost-overruns), 
was "well within" this outer perimeter. Id. 

70. Id. at 2705-06. 

71 . 102 S. Ct. 2727 (1982). The action alleged that the two close aides of President Nixon, 
Bryce Harlow and Alexander Butterfield, conspired with the President to unlawfully discharge 

72. 408 U.S. 606 (1972). 

73. 102 S. Ct. at 2734-35. 

74. Id. 

75. Id. at 2738. 


was not clearly established, an official could not reasonably be expected to 
anticipate subsequent legal developments, nor could he fairly be said to 
'know' that the law forbade conduct not previously identified as unlawful. 
Until this threshold immunity question is resolved, discovery should not be 
allowed. If the law was clearly established, the immunity defense ordinarily 
should fail, since a reasonably competent public official should know the 
law governing his conduct. Nevertheless, if the official pleading the defense 
claims extraordinary circumstances and can prove that he neither knew nor 
should have known of the relevant legal standard, the defense should be 
sustained. But again, the defense would turn primarily on objective factors. ^^ 

The most recent case on governmental immunity to reach the Supreme Court 
was remanded for further consideration in light of this decisicui>^^ 

— The Case Law's Matrix of Issues 

The foregoing essentially chronological description of the case law can be 
usefully complimented by a review organized according to the most significant 
immunity law issues. Any analysis of the matrix of issues presented by the case 
law on the immunity doctrine will inevitably be less than crystal clear, as the 
decisions in this area have not been very consistent, especially in recent years. 

a. Source of Immunity 

In determining the degree of immunity to be granted to various government 
officials, the courts have looked to a number of sources for guidance. The 
paramount consideration has been that of public policy — what degree of im- 
munity will best serve societal interests — but the courts have not always con- 
sidered all of the competing policy considerations and they have examined other 
factors as well. For example, with regard to judicial immunity, the Supreme 
Court considered the English common-law rule, as well as policy considera- 
tions.^^ In determining the appropriateness of a similar blanket immunity for 
federal legislators, the Court found, of course, the shield from liability to be 
rooted in the Speech or Debate Clause of the Constitution.^^ And the scope of 
Presidential immunity was considered by Justice Powell to be "a functionally 
mandated incident of the President's unique office, rooted in the constitutional 
tradition of the separation of powers and supported by our history, "^^ and was 
justifiable on policy grounds. The vigorous dissent by Justice White, on the 

76. Id. at 2739. 

77. Velde v. National Black Police Ass'n., 102 S. Ct. 3503 (1982). This case is particularly 
significant in that it involves the immunity issue in the context of § 1983 actions involving federal 
grants. See infra Chapter V, §A. 

78. Bradley v. Fisher, 80 U.S. (13 Wall.) 335, 347-49 (1872). The purported necessity of 
absolute judicial freedom was the paramount consideration of the Court, however. 

79. U.S. Const, art. I, § 6, cl. 1. The Court has held that "legislators engaged 'in the sphere 
of legitimate legislative activity' . . . should be protected not only from the consequences of liti- 
gation's results but also from the burden of defending themselves." Dombrowski v. Eastland, 387 
U.S. 82, 85 (1967); see also Kilboum v. Thompson, 103 U.S., 168, 204 (1881). 

80. Nixon v. Fitzgerald, 102 S. Ct. 2690, 2701 (1982). 


Other hand, in strongly rejecting the constitutional underpinnings of the decision 
characterized the majority decision as "almost wholly a policy choice [and] a 
choice that is without substantial support."^' 

For lower executive officials, the Court's so-called "functional" approach 
has been based, in part, on public policy grounds. The most important of the 
grounds favoring immunity is the fear that broad liability to suit will deter officials 
from principled and forceful decisionmaking^^ and the belief that such broad 
liability for officials who must exercise discretion is unjust. ^^ Balanced against 
these considerations is the need for a remedy for those who have been wrongly 
injured, and the policy of deterring government officials from carelessly or 
intentionally taking actions that will infringe citizens' rights. 

b. Degree of Immunity Available 

There are generally two basic categories of immunity to be provided for 
government officials — absolute and qualified good-faith immunity. Absolute 
immunity is available to judges,^'* federal legislators,^^ prosecutors,^^ and the 
President. ^^ In addition, legislative aides are covered by this blanket protection, ^^ 
as are agency officials who perform functions analogous to those of a prosecutor 
or judge. ^^ The grant of absolute immunity does not mean that these officials 
may never be sued for money damages. For example, a judge is not immune 
from suit when he has injured someone in a case where the court clearly has no 
jurisdiction,^^ and a federal legislator is only protected for actions taken "in the 
sphere of legitimate legislative activity. "^^ Similarly, the President receives 
absolute immunity only for actions taken within the outer perimeter of his official 

For executive officials other than the President, prosecutors, and those who 
perform analogous functions to those of a prosecutor or judge, only qualified 
immunity is available as a defense to a suit for money damages. Even this 
qualified defense may be available only for discretionary actions. As the Court 
explained in Scheuer v. Rhodes, the degree of immunity varies depending upon 
"the scope of discretion and responsibilities of the office and all the circumstances 
as they reasonably appeared at the time of the action on which liability is sought 

81. id.dxim. 

82. See, e.g.. Wood v. Strickland, 420 U.S. 308, 319-20 (1975); Pierson v. Ray, 386 U.S. 
547, 554 (1967). 

83. See, e.g., Scheuer v. Rhodes, supra note 51, at 240. 

84. See Bradley v. Fisher, 80 U.S. (13 Wall.) 335 (1872). 

85. See Dombrowski v. Eastland, supra note 79; Kilboum v. Thompson, supra note 79. 

86. See Imbler v. Pachtman, 424 U.S. 409 (1976); Tyler v. Witkowski, 511 F.2d 449 (7th 
Cir. 1975). 

87. Nixon v. Fitzgerald, supra note 67. 

88. Gravel v. United States, 408 U.S. 478 (1978). 

89. See, e.g., Butz v. Economou, 438 U.S. 478 (1978). 

90. Bradley v. Fisher, supra note 84. 

91. Dombrowski v. Eastland, supra note 79, at 85. 

92. Nixon v. Fitzgerald, supra note 67. 


to be based. "^^ This qualified good-faith defense applies to local, state and 
federal executive officials. It is applicable to governors, ^"^ Cabinet officials, ^^ 
and presidential aides, ^^ as well as to lower-ranking officials. The Court has left 
open the possibility that certain high-ranking federal officials may be entitled to 
absolute immunity if public policy would require such a blanket protection. ^^ 
But these officials would bear the burden of showing that public policy requires 
a grant of absolute immunity for them.^^ 

c. Discretionary and Mandatory^ Action 

The Court has indicated in its decisions over the past twenty-five years that 
officials with greater discretionary authority are to be protected to a greater 
degree by the immunity defense than those exercising less discretion in their 
actions. This result, it has been noted, is perverse in that it is officials who 
exercise the least amount of discretionary authority who are most prone to damage 
actions and may be the most likely to react to the risk of suit by performing 
their duties in a manner designed to avoid suit rather than promote the public 
interest, ^^ as the Court stated in Barr v. Matteo: 

To be sure, the occasions upon which the acts of the head of an executive 
department will be protected by the privilege are doubtless far broader than 
in the case of an officer with less sweeping functions. But that is because 
the higher the post, the broader the range of responsibilities and duties, and 
the wider the scope of discretion it entails. ^^ 

In Scheuer v. Rhodes, the Court, in granting a qualified immunity for both 
the governor of a state and lower executive officials, decided that "the variation 
[in immunity is] dependent upon the scope of discretion and responsibilities of 
the office and all the circumstances as they reasonably appeared at the time of 
the action on which liability is sought to be based. "'^' In its remand, the Court 
directed the district court to determine whether the officers acted within the range 
of discretion permitted for the holders of these offices; it also indicated that the 
governor's office would permit the exercise of more discretion than would be 
the case for lower officers. '^^ 

93. 416 U.S. at 247. 

94. Id. at 247-48. 

95. Butz V. Economou, supra note 61. 

96. Harlow v. Fitzgerald, supra note 71. 

97. For example, the Court indicated, "[f]or aides entrusted with discretionary authority in 
such sensitive areas as national security or foreign policy, absolute immunity might well be justified 
to protect the unhesitating performance of functions vital to the national interest." Id. at 2735. 

98. Id. at 2735-36. See also Butz v. Economou, supra note 61. 

99. Schuck, Suing our Servants: The Court, Congress, and the Liability of Public Officials 
for Damages, 1980 S. Ct. Rev. 281, 293-95, 299-300. 

100. 360 U.S. 564, 573 (1959). The defendant, the acting director of the Office of Rent 
Stabilization, exercised discretion in his duties. 

101. 416 U.S. 232,247(1974). 

102. Id. at 250. The Court stated, "[i]n short, since the options which a chief executive and 
his principal subordinates must consider are far broader and far more subtle than those made by 
officials with less responsibility, the range of discretion must be comparably broad." Id. at 247. 


In Butz V. Economou,^^^ the Court rejected absolute immunity for Cabinet 
officials, despite its recognition that such high officials are required to constantly 
exercise their discretion. Balanced against the need to protect officials who are 
required to exercise discretion was the Court's fear that "the greater power of 
such officials affords a greater potential for a regime of lawless conduct. "^^ 
Thus, only qualified immunity was appropriate. These same considerations were 
reiterated in awarding only a qualified immunity to high White House officials, 
who also must exercise their discretion in day-to-day decisionmaking. ^^^ 

Where an official performs no discretionary functions and engages only in 
ministerial duties, the law is well-settled that no immunity exists for torts com- 
mitted. This concept has been the common law and has remained virtually 
unchanged over the years. The greater question lies in determining what functions 
are discretionary.'^^ 

d. Scope of the Defendant' s Authority 

Another issue is whether a defendant official has acted within the scope of 
his authority. Where he has acted within that scope, the official is protected by 
whatever type of immunity is applicable — otherwise, he is not. Perhaps the wide 
variety of situations in which the immunity defense can be raised has inhibited 
the Court from articulating rules of general applicability and the cases are often 
remanded for a determination of that factual issue. '^^ 

The standard for the determination has been phrased in various ways, such 
as "within the outer perimeter of [the official's] line of duty,"'^^ and "acts 
performed in the course of official conduct. "'^^ In the few cases where the 
Supreme Court has made this factual determination, the accompanying analysis 
and discussion have not been very helpful. For example, in holding that the 
acting director of the Office of Rent Stabilization acted within the scope of his 
authority in stating in a press release that he planned to suspend certain employees 
for conduct for which the agency had been criticized, the Court merely stated 
that it would be "an unduly restrictive view of the scope of the duties of a 
policy-making executive official to hold that a public statement of agency policy 
in respect to matters of wide public interest and concern is not action in the line 
of duty." The Court added that the fact that the official was not required by law 
or by his superiors to make the statement was not a factor to be considered, 
since he was of policy-making rank, acting at a level of government where the 
concept of duty encompasses the sound exercise of discretionary authority. '^^ 

103. 438 U.S. 478 (1978). 

104. Id. at 506. 

105. Harlow v. Fitzgerald, 102 S. Ct. 2727 (1982). 

106. K. Davis, Administrative Law Treatise §§ 26.15-16 (Supp. 1982). 

107. See, e.g., Scheuer v. Rhodes, supra note 51, at 502. 

108. Barr v. Matteo, supra note 45, at 575. 

109. Scheuer v. Rhodes, supra note 51, at 248. 

1 10. Barr v. Matteo, supra note 45, at 575. 


Similarly, President Nixon was considered to be acting within the outer 
perimeter of the duties of his office in ordering a departmental reorganization 
in the Air Force, in which the job of the plaintiff (who had testified before 
Congress on military cost-overruns) was eliminated. In so ruling, the Court only 
stated that finding otherwise would subject the President to trial on numerous 
allegations, and would deprive the absolute immunity of its intended effect. ^^' 
Thus, in those few cases where the Supreme Court has addressed the question 
of the scope of a defendant official's authority, it has allowed a broad scope for 
officials that must exercise discretion. ^'^ 

e. Piercing the Immunity Defense 

As discussed above, neither an absolute nor a qualified immunity is available 
to a defendant official where the actions which form the basis of the suit were 
not within the outer perimeter of the official's authority. Thus, such a threshold 
demonstration by the plaintiff could preclude the immunity defense. However, 
this is often hard to accomplish. With very high officials such as the President, 
it may be almost impossible to argue that a particular action was not within the 
outer reach of the scope of their office.''^ 

The qualified good-faith immunity is an affirmative defense that must be 
pleaded by a defendant official not entitled to absolute immunity. '^"^ By dem- 
onstrating that the official did not have a good-faith belief that he was not violating 
the plaintiff's rights, the plaintiff may successfully attack this defense. Previ- 
ously, this good-faith test contained both objective and subjective elements, '^^ 
but in its 1982 decision in Harlow v. Fitzgerald, the Supreme Court decided 
that inquiries into whether a defendant official acted with a malicious intention 
only served to protect insubstantial claims from summary judgment. It therefore 
eliminated this subjective element from the test. Now only a showing that the 
official ''knew or reasonably should have known that the action he took within 
his sphere of official responsibility would violate the constitutional rights of the 
[plaintiff]" will pierce the qualified immunity defense. ^^^ 

Furthermore, the defense will not fail if the law upon which the plaintiff 
relies was not clearly established at the time the official acted. In this circum- 
stance, the Court reasoned, the official could not fairly be said to "know" that 
the law forbade the actions which he took. But if the law was clearly established, 
the defense will ordinarily fail; only a showing by the official that he neither 


111. Nixon V. Fitzgerald, supra note 67, at 2705. 

112. A related issue is whether official immunity should be available for action under color 
of official title. See infra Chapter IV, § B. 

113. In Nixon v. Fitzgerald, for example, it was argued without success that the alleged 
misconduct by the President involved illegal interference with a congressional witness which was 
outside the scope of the President's authority. See, e.g., ^rxQi Amicus Curiae of Senator Orrin G. 
Hatch, et al. at 20, Nixon v. Fitzgerald, 102 S. Ct. 2690 (1982). 

114. Gomez v. Toledo, 446 U.S. 635, 640 (1980). 

115. Wood V. Strickland, supra note 55, at 322. 

116. Harlow v. Fitzgerald, supra note 71, at 2737 (emphasis in the original). 


knew nor should have known of the relevant legal standard will allow his defense 
to be sustained. ^'^ 

It remains to be seen, however, whether the new good-faith standard will 
have the intended effect of facilitating the early dismissal of suits. It may be, 
for example, that the so-called objective test contains in many cases subjective 
issues. The realm of other difficult issues that could frustrate early dispositions 
is as vast as the ingenuity of counsel. 

— An Evaluation 

The case law establishes a confusing and often unsatisfactory array of rules 
for official liability and immunity. There is virtually no evidence that these rules 
accomplish the social objectives they are purported to advance such as compen- 
sation, deterrence and encouraging desirable official conduct. 

But the judicial process, by its very nature is ill-suited to make the type of 
comprehensive policy decisions that many believe are needed in this area.^*^ 
The courts depend on individual cases which arise serendipitously, and they are 
decided one by one, rather than in a cohesive manner most conducive to general 
policy. Moreover, courts are often hampered and insulated from the acquisition 
of the type of factual record that would inform broad policy choices. 

Criticism of the courts' liability and immunity rules comes easily in such 
circumstances. But the courts are only filling vacuums not grabbing turf. It seems 
obvious that initiatives for changing the present civil sanction system's primary 
reliance on the private damage action must come from the legislative arena with 
a strong assist from the executive branch. One commentor suggests that this 
process, now well underway, is actually promoted by the difficulties created by 
the case law: 

[T]he [Supreme] Court's approach can be sensibly viewed as a heroic effort 
to fashion a just system of incentives and deterrents in the face of chronic 
legislative inaction. The Court, in this view, has fashioned a respectable 
legal artifact given the crudeness of the tools available to it. Moreover, by 
magnifying the anomalies of this ineffective and often unjust system, the 
Court's effort could stimulate a rationalization and strengthening of that 

B. Existing Forms of Public Enterprise Liability 

In response to the increased threat of personal liability caused by the several 
factors discussed above, various attempts have been made at all levels of gov- 
ernment to shift this risk elsewhere. '^° The following sections will consider 
several of the existing arrangements for shifting damages from personal liability 
to governmental or enterprise liability at the federal and state levels, including: 

117. Id. ai 2139. 

118. See infra Chapters IV and V. 

1 19. Schuck, supra note 99, at 286-87 (footnotes omitted). 

120. See infra Chapters IV and V, and Appendix B. 


(1) liability insurance for the individual official, (2) governmental indemnifi- 
cation of the official, and (3) direct liability of the governmental enterprise for 
the actions of its employees. 

1 . Federal Enterprise Liability ' ^ ' 

— Insurance 

One method available for shifting the risk of liability away from the indi- 
vidual official is the utilization of liability insurance. '^^ In the private sector 
individuals frequently make use of liability insurance to avoid large financial 
losses caused by accidents, fire, theft, etc. Government employees could also 
utilize liability insurance to protect against the risk of a large adverse judgment. 
The employee could purchase such insurance individually or be covered instead 
by a group program purchased by his employer. By paying a premium to an 
insurance company, the risk of liability is shared with the third-party insurer. '^^ 
The individual employee is thus freer to operate without the threat of personal 
liability hanging over his head. 

Federal law generally does not authorize the purchase of insurance against 
the personal liability of officials. '^"^ One reason for this limited use of insurance 
as a means of shifting liability to the enterprise is that such a program might be 
contrary to public policy because it reduces the deterrent value of a personal 
judgment against the government official. '^^ To the extent that insurance is 

121. There are three forms of enterprise liability. Under a system of liability insurance the 
insurer receives premium payments and pays the plaintiff for loss. Under a system of indemnification, 
the enterprise reimburses the individual defendant who pays off the plaintiff. In contrast, under a 
system direct enterprise liability, the government itself would pay the plaintiff directly. 

122. Mashaw, supra note 36, at 8. 

123. A federal agency usually will decide to assume the risk of its own liability thereby 
acting, in effect, as a self-insurer. According to Federal Procurement Regulations: 

Ordinarily, it is the policy of the Government not to insure its own risks. In the absence 
of specific statutory authority for the payment of insurance premiums, appropriated moneys 
of the United States generally are not regarded as available for that purpose. 

41 C.F.R. §1-10.301 (I98I). The Comptroller General has stated: 

It is the established policy of the Government to be a self-insurer of its own risks of loss 
on the theory that size of the Government resources permits it to do so. . . . Comp. Gen. 
Dec. B-168106, 74-2 CPD 113 (1974). 

124. There are limited areas where the Government will purchase insurance which would 
cover negligence for which its employees would be otherwise personally liable. These include insuring 
the risk of loss for the first one hundred dollars of property damage in auto accidents involving 
rented cars, and also damage for rented aircraft, and rented buses. See Comp. Gen. Dec. B-168106, 
74-2 CPD 113 (1974). Certain medical personnel obtain liability insurance. See, e.g., 42 U.S.C. §§ 
2458a and 2458b. Such insurance may be distinguished from indemnification of federal officials 
which is available on a limited basis. Federal officials have difficulty purchasing liability insurance 
on their own. See, Amendments to the Federal Tort Claims Act: Hearings before the Subcommittee 
on Administrative Law and Governmental Relations of the House Committee on the Judiciary, 96th 
Cong., 1st Sess. 88 (1979) (statement of Ordway Burden, Law Enforcement Assistance Foundation). 

1 25 . See Schuck, supra note 99, at 332 and n. 1 79; Cass, Damage Suits Against Public Officers, 
129 U. Pa. L. Rev. 1110, 1160-64 (1981). 




effective, it may so fully reduce the incentive to avoid liability-producing activity 
that underdeterrence becomes a problem. '^^ 

A second reason for the limited reliance on insurance is the costs associated 
with dealing with an intermediary (the insurance company) which would not 
exist in a system of direct enterprise liability. '^^ 

— Indemnification 

Another method of shifting the risk of liability away from the individual 
official is indemnification. Under a statutory system of indemnification, the 
government itself would be required to pay any adverse judgment suffered by 
the individual official, including the cost of defending the action. At the present 
time, there is no general statutory provisions for indemnification of federal 
employees, even where the liability clearly arose within the scope of their em- 

The major reason why indemnification is not frequently used as a substitute 
for personal liability is the potential for a conflict of interest between the de- 
fendant enterprise and the defendant employee. For example, a defendant em- 
ployee may defend a tort action by claiming that the enterprise was solely 
responsible for the alleged tortious conduct. In such a situation, the enterprise 
would, in effect, reimburse the employee for the employee's efforts to place 
liability on the enterprise. Another reason why indemnification is not widely 
used is that it results in the same form and amount of compensation as direct 
enterprise liability, but involves an extra step in paying out a claim and the extra 
administrative costs that go along with that step. Finally, indemnification does 
not eliminate the necessity for officials to participate in and defend lawsuits. 

— Direct Government or Enterprise Liability 

Prior to the enactment of the Federal Tort Claims Act (FTC A), damage 
suits against the United States arising from common-law torts were generally 
barred by sovereign immunity. '^^ Created in England and developed in the United 
States, sovereign immunity stands for the proposition that the sovereign cannot 
be sued without its consent. '^° The doctrine of sovereign immunity has no explicit 

126. Cass, supra note 125, at 1 160. Proponents of enterprise liability claim that such a system 
is necessary to overcome the problem of overdeterrence. 

127. As the Comptroller General explains: "Self insurance [by the Government] will affect 
in a survey of those items of cost and profit that would normally be included in the premiums charged 
by private insurers." Comp. Gen. Dec. supra note 123. 

128. Schuck, supra note 99, at 333; Hearings on S. 1775 Before the Subcommittee on Agency 
Administration of the Senate Committee on the Judiciary, 97th Cong. 1st Sess. (1981) (statement 
of Edward C. Schmults). Schuck reports that there are only two exceptions in which indemnification 
is specifically authorized by statute: "for actions based on 'anything done' in enforcement of the 
Internal Revenue Code (26 U.S.C. § 7423) and for malpractice suits against medical personnel of 
several named agencies (See 10 U.S.C. § 1089(f), 22 U.S.C. § 817(f), 38 U.S.C. § 4116(e), and 
42 U.S.C. §§ 233(f) and 2458a(b))." Schuck, supra note 99, at 333 n.l80. 

1 29. Comment, Scope of the Discretionary Function Exception Under the Federal Tort Claims 
Act, 67 Geo. L.J. 879, 882 (1979). 

130. Basso V. United States, 239 U.S. 602 (I9I6). 


constitutional or statutory basis. Rather, it emerged as a result of judge-made 
law.'^' Chief Justice Marshall, in Cohens v. Virginia stated: 

"The universally received opinion is, that no suit can be commenced or 
prosecuted against the United States. . . ."'^^ 

Because of the sovereign immunity bar, one of the only remedies available 
to an individual tortiously aggrieved by an agent of the government was a "private 
bill" from Congress allowing recovery against the United States government. '-^-^ 
The growing number and burden of private bills'-^'* and mounting hostility to 
sovereign immunity eventually led Congress to pass the FTCA and thus create 
a simplified recovery procedure for those injured by governmental negligence. 
The FTCA waived the United States' immunity from suit for claims arising out 
of the "negligent" or "wrongful" conduct of government employees acting 
"within the scope of employment." 28 U.S.C. § 1346(b) (1976) provides: 

[T]he district courts . . . shall have exclusive jurisdiction of civil actions 
on claims against the United States, for money damages ... for injury . . . 
caused by the negligent or wrongful act or omission of any employee of 
the Government while acting within the scope of his office or employment, 
under circumstances where the United States, if a private person, would be 
liable to the claimant in accordance with the law of the place where the act 
or omission occurred. 

Later amendments to the FTCA also waived the United States' immunity in suits 
arising from the "intentional torts" of its employees, such as assault, battery, 
false imprisonment, false arrest, abuse of process and malicious prosecution 
when committed by federal investigative or law enforcement officers. '^-^ 

The FTCA as it currently reads, however, leaves the immunity of the United 
States government intact in several important areas. First, the government cannot 
be held liable for the performance or non -performance of a discretionary function 
or duty. 28 U.S.C. § 2680 provides: 

[T]his title shall not apply to — 

(a) Any claim based upon ... the performance or the failure to 
exercise or perform a discretionary function or duty on the part of a federal 
agency or an employee of the Government, whether or not the discretion 
involved be abused. 

Second, it appears that the United States cannot be held liable in any tort action 
based on a constitutional claim. This exception to the FTCA stems from the fact 

131. Comment, Constitutional Tort Remedies: A Proposed Amendment to the Federal Tort 
Claims Act, 42 Conn. L. Rev. 492, 499 (1980). 

132. 19 U.S. (6 Wheat.) 264 (1821). 

133. 67 Geo. L.J. 879, 883 (1979). An injured plaintiff could also bring an application for 
administrative settlement. Id. 

134. See Dalehite v. United States, 346 U.S. 15, 24-25 n.9 (1953); Symposium, 42 Law & 
Contemp. Prob. at 86-87. 

135. Intentional Tort Amendment Act of 1974, Pub. L. No. 93-253, § 2, 88 Stat. 50 (codified 
at 28 U.S.C. § 2680(h)). 


that under 28 U.S.C. § 1346(b), the United States can only be liable where a 
private person would be liable under the state law. Since the constitutional tort 
is based on federal law, the argument follows that the United States cannot be 
liable for constitutional torts under the FTCA and thus retains its immunity. '^^ 
Finally, the United States generally is able to assert the immunities and defenses 
of its employees as a defense in an action covered by the FTCA. ^^'^ For example, 
under present law the United States cannot be held liable for any tort committed 
by a government employee who was acting in good faith. Although the FTCA 
is the predominant form of enterprise liability at the federal level, there are other 
statutes that have authorized suits against the United States for wrong of its 
employees. These include the Swine Flu Immunization Act,'^^ and the Teton 
Dam Act.'^^ 

2. State Enterprise Liability 

The extent to which state governments provide for the shifting of the risk 
of liability away from the government official or employee varies from state to 
state. '"^^ Many states now provide for the protection of state government officials 
and employees in the form of free defense counsel, indemnification or insur- 
ance. '"*' In addition, some states extend these protections to officials and em- 
ployees of a state's political subdivisions. '"^^ However, the availability of these 
protections in every case is contingent upon that official's having acted in good 
faith. '4^ 

While there is clearly a trend today among state governments to adopt some 
form of governmental assumption of liability, a significant number of states limit 
the degree to which the state will indemnify an official, '"^"^ while a few states 
still have no formal or extensive system for direct governmental liability. '"^^ 
Moreover, in a limited number of states, sovereign immunity is still the rule.^"^^ 

136. Bell, Proposed Amendments to the Federal Tort Claims Act, 16 Harv. J. on Legis. 1, 
4 (1979). 

137. Norton v. United States, 581 F.2d 390 (4th Cir. 1978). 

138. Pub. L. No. 94-380, 90 Stat. 1 1 13 (1976) (codified at 42 U.S.C. § 247b). The liability 
provisions were subsequently repealed by Pub. L. No. 95-626, 92 Stat. 3551 (1978). 

139. Pub. L. No. 94-400, 90 Stat. 1211 (1976). 

140. Recent developments are discussed in Chapter V and Appendix B at length. 

141. See, e.g., Ariz. Rev. Stat. § 41-621, etseq.; Nev. Rev. Stat. §§ 41.0339 and 41.0349. 
See also Schuck, supra note 99, at 334. 

142. See, e.g., Cal. Gov't Code § 995. 

143. The good-faith aspect of state protection of officials and employees means that the 
officials must be found to have acted within the "[c]ourse and scope of employment." See, e.g., 
Mass. Gen. Laws ch. 258, § 2; Minn. Stat. Ann. § 3.736(9). 

144. See, e.g., Ky. Rev. Stat. § 44.070. 

145. See letters from Wilson L. Condon, Attorney General of Alaska and Jeffrey P. Hallem, 
Assistant Attorney General of South Dakota, to Thomas J. Madden, contained in Appendix B. 

146. See letter from Robert O. Welford, Attorney General of North Dakota, to Thomas J. 
Madden, contained in Appendix B. 


C. Administrative Controls Over Federal Officials^"^^ 

The existing system of administrative disciplinary controls on officials con- 
stitutes a third method for dealing with governmental wrongs. The procedures 
which are available vary greatly depending on the official position held by the 
alleged wrongdoer. For example, career civil servants are subject to disciplinary 
proceedings with established, explicit, substantive and procedural norms. ''^^ These 
internal disciplinary proceedings are of particular interest because career civil 
servants represent the vast majority of government officials and are the ones 
most often sued. In contrast, administrative controls are not generally available 
for elected officials, political appointees and former officials. These latter types 
of officials are, however, sensitive to external controls such as adverse publicity, 
congressional oversight and judicial review. '"^^ The application of both internal 
and external controls to former officials of all types poses special problems for 

It is important to recognize that the fundamental purpose of the existing 
administrative controls is to promote administrative efficiency and not to deter 
official wrongdoing that harms individual members of the public. The principal 
legislative mechanism permitting agencies to discipline officials provides for 
sanctions that will "promote the efficiency of the service." '^^ As Supreme Court 
Justice Powell recognized in Arnett v. Kennedy. "[T]he Government, as an 
employer, must have wide discretion and control over the management of its 
personnel and internal affairs. This includes the prerogative to remove employees 
whose conduct hinders efficient operation and to do so with dispatch. "^^' 

There is no general agreement on whether the existing administrative dis- 
ciplinary system adequately satisfies either the goals of efficiency or deterrence. 
In light of the obvious elusiveness of quantifying the system's effectiveness in 
achieving these goals, it is hardly surprising that there are few promising avenues 
for evaluating the system, much less any substantial empirical evidence of the 
performance of administrative controls. '^^ 

147. This section will discuss existing administrative sanctions only at the federal level. It is 
both infeasible and unnecessary to describe the myriad administrative controls in each of the fifty 
states and local sanctions as well. 

148. 5^e, t-.g., 5 U.S.C. §§ 7101-7901. 

149. External controls are examined infra at pp. 3-64 through 3-70. 

150. 5 U.S.C. § 7503(a). 

151. 416 U.S. 134. 168 (1974). 

152. In developing a model for conducting the necessary evaluation a starting point would 
be the recent literature on the theory and historical development of administrative agencies. See, 
e.g., Stewart, Reformation of Administrative Law, 88 Harv. L. Rev. 1667, 1677-78 (1975). Ack- 
erman and Hassler, Beyond the New Deal, 89 Yale L.J. 1466, 1471-72 (1980); see also 1 K. Davis, 
Administrative Law Treatise §§ 1.6-1.9 (1978). 

Several commentators suggest, however, that the administrative disciplinary system does little 
to improve the efficiency of the service noting that inefficiency is not a grounds for dismissal or 
susf)ension in Civil Service adverse action. See Mashaw. supra note 36; Merrill, supra; R. Vaughn, 
The Spoiled Systems, supra. It has even been suggested that civil service regulations, in general, 
limit the government's ability to reward good, as well as punish poor, job performance, thus removing 


1 . Individual Sanctions 

— Elected Officials 

The President and the Vice President are not subject to the administrative 
sanctions imposed on other officials from within the executive branch of gov- 
ernment. They are not subject to civil service disciplinary procedures, and, under 
the Constitution, they can only be removed from office by the electorate or after 
impeachment. Of course, at the state level where there are often numerous elected 
executive officials, the problems of disciplining these officials are more signif- 

— Political Appointees 

Political appointees are subject to different administrative sanctions than 
are civil servants who are members of the "competitive service." '^^ They may 
be treated more harshly than other civil servants, since they are not protected 
by the statutory and regulatory procedures designed to protect civil servants from 
arbitrary treatment. '^"^ Political appointees may be removed at the will of the 
President. This removal power is said to be implied in the appointment power 
set out in Article 11 of the Constitution. The extreme sanction of removal, 
however, may not be too meaningful since the President might want to retain 
an otherwise valuable official. While the President may unilaterally remove 
certain executive officials, congressional concurrence may be required to support 
the removal of appointed officials of independent agencies or quasi-legislative 
or judicial officials. '^^ 

Excepted Service 

The President has authority to appoint a limited number of persons to 
positions specifically excepted from the competitive service, as defined below. 
These include a number of sub-cabinet level appointments of individuals believed 

incentives for employees to engage in desirable behavior. Cass, supra note 125. One authority has 
noted that he was unaware of any case "in which the most extreme pecuniary sanctions available — 
loss of a government job or demotion — have been imposed for mere negligence." Mashaw, supra 
note 36, at 23. 

153. According to the Lloyd-LaFollete Act, 5 U.S.C. §§ 2101 et seq., those persons whose 
appointments are made by nomination for confirmation by the Senate are not part of the "competitive 
service." Under Article II, § 2 of the Constitution, certain appointments made by the President must 
be approved by two-thirds of the Senate. Officials appointed to these positions are those whom this 
report refers to as political appointees. 

154. Id. 

155. Compare Myers v. United States, 272 U.S. 52 (1926) (President may remove postmaster 
without congressional approval) with Humphrey's Executor v. United States, 295 U.S. 602 (1935) 
(President may not remove FTC Commissioner without congressional approval) and Wiener v. United 
States, 357 U.S. 349 (1958) (President may not remove War Claims Commissioner without congres- 
sional consent). 


by the President and his officers to be desirable in order to carry out poHcy and 
poHtical objectives. They are referred to as ''Schedule C" officials. '^^ 

— Competitive Service 

The competitive service is statutorily defined in the Lloyd-LaFollete Act: '^^ 

(a) The 'competitive service' consists of — 

(1) all civil service positions in the executive branch, except — 

(A) positions which are specifically exempted from the competitive 
service by or under statute; 

(B) positions to which appointments are made by nomination for 
confirmation by the Senate, unless the Senate otherwise directs; 

(C) positions in the Senior Executive Service. '^^ 

a. Reduction in Grade or Removal by the Agency for Unacceptable 


Federal civil servants are generally subject to performance evaluations.'^^ 
Each agency must develop a performance appraisal system to evaluate employees 
periodically and provide a basis for promoting, reassigning, and demoting them. 
The system is meant to "recogniz[e] and reward[ ] employees whose perfor- 
mance so warrants," and "assist[ ] employees in improving unacceptable per- 
formance."'^ The applicable statutory and regulatory provisions do not define 
in detail what conduct is so unsatisfactory as to warrant disciplinary action. Gross 
abuses of power or constitutional violations may fall within the definition.'^' 

Subject to notice and hearing requirements, "an employee may be reduced 
in grade or removed at any time during the performance appraisal cycle that the 
employee's performance in one or more critical elements of the job becomes 
unacceptable."'^^ A "critical element" is "a component of an employee's job 
that is of sufficient importance that performance below the minimum standard 
established by management requires remedial action" such as removal or a grade 

156. 5 C.F.R. § 6.2. Schedule C positions are those of a confidential or policy-determining 
character. Except as otherwise may be required by statute. Civil Service rules and regulations do 
not apply to removals from positions in Schedule C or from positions excepted from the competitive 
service by statute. Exec. Order No. 10577, 5 U.S.C. § 3301 note. 

157. "Civil service" is defined in the statute to include "all appointive positions in the 
executive, judicial, and legislative branches of the Government of the United States, except positions 
in the uniformed services." 5 U.S.C. § 2101(1). 

158. 5 U.S.C. § 2102. Although the statute here distinguishes between the "competitive 
service" and the "Senior Executive Service," members of these two groups are generally subject 
to the same kinds of administrative sanctions and receive the same statutory and regulatory protection. 
The statute also provides that even those whose appointments are subject to Senate confirmation 
may be specifically included in the competitive services by statute. 5 U.S.C. § 2102(b). 

159. The details of which federal employees are subject to this sort of evaluation are set out 
in 5 U.S.C. § 4301, 5 C.F.R. § 432.201(b), 5 C.F.R. § 432.201(c), and statutes and regulations 
cited therein. 

160. 5 U.S.C. § 4302. 

161. See generally 5 C . F. R . Part 432 . 

162. 5 C.F.R. § 432.203(a). 


reduction, even without regard to the employee's performance on other com- 
ponents of the job.'^'' However, a reduction in grade or removal may be based 
only on those instances of unacceptable conduct which occurred within the past 

Disciplinary actions for unsatisfactory performance must be initiated by the 
employee's own agency. The statute does not provide for either a private right 
of action by members of the public, or for direct prosecution by an outside 
agency such as the Office of Personnel Management (0PM). '^^ However, mem- 
bers of the public, or other agencies might of course bring an employee's actions 
to the attention of his or her agency in an attempt to precipitate disciplinary 
proceedings. They might also try to employ publicity to spur the agency into 

b. Disciplinary Sanctions Imposed By the Agency to "Promote the 

Ejficiency of the Service" 

Agencies may remove, suspend, reduce in grade, reduce in salary, or fur- 
lough for thirty days or less, employees, when such measures would "promote 
the efficiency of the service. "'^^ These sanctions are brought into play by the 
employee's own agency, rather than by a private right of action, and must be 
accompanied by certain procedural protections. 

163. 5 C.F.R. § 430.202(e). An employee whose reduction in grade or removal is proposed 
because of an unsatisfactory performance is entitled to thirty days notice of the charges against him 
or her, representation by an attorney, an opportunity to answer the charges orally and in writing, 
and a written decision. 5 U.S.C. § 4303(b). An employee who is reduced in grade or removed 
because of inadequate performance may appeal the agency's decision to the Merit Systems Protection 
Board, a federal body dealing with disciplinary hearings in most of the agencies. 5 U.S.C. § 4303(e). 
The Merit Systems Protection Board and this appeal mechanism are described in detail in Rescuing 
Riffs: An Overview of the Merit Systems Protection Board 6 District Lawyer 32-35 (1982). In cases 
involving a removal, the Board itself must hear the appeal. When only a demotion is involved, the 
Board may appoint an administrative law judge to hear the appeal. The agency's decision may be 
sustained only if it is based on "substantial evidence." 5 U.S.C. § 7701(c)(1)(A). Under certain 
circumstances the Office of Personnel Management may also participate in such appeals. 5 U.S.C. 
§ 7703(d). Finally, "[a]ny employee or applicant for employment adversely affected or aggrieved 
by a final order or decision of the Merit Systems Protection Board may obtain judicial review of 
the order or decision." 5 U.S.C. § 7703(a)(1). The procedures, jurisdiction, and standards appropriate 
to such appeals are set out in 5 U.S.C. § 7703. 

The courts are frequently deferential to decisions made by the Merit Systems Protection Board. 
See, e.g., Gueory v. Hampton, 510 F.2d 1222, 1225 (D.C. Cir. 1974); Mendelson v. Macy, 356 
F.2d 796 (D.C. Cir. 1966). 

164. 5 U.S.C. § 4303(c)(2)(A). 

165. The 0PM can, however, within one year of an employee's appointment, bring a direct 
action against employees when discipline is necessary "to promote the efficiency of the agency." 
The Office of Special Counsel can in appropriate circumstances prosecute a disciplinary action. 5 
U.S.C. §§ 1201-09. 

166. See 5 U.S.C. §§ 7503, 7513, 7543; 5 C.F.R. §§ 752.101-752.606. This group of 
sanctions is referred to as "adverse actions." 


The phrase "promote the efficiency of the service" is broad. '^^ Neverthe- 
less, OPM regulations defining how agencies shall determine whether a sanction 
is appropriate illustrate the type of misconduct that is meant to be subject to 
disciplinary actions. '^^ It is apparent that these regulations are primarily directed 
to promoting administrative efficiency and not toward deterring official wrongs 
which result in harm to individual members of the public. 

Courts have upheld dismissals of employees for a broad range of conduct 
such as leaving a job without permission for eleven days,'^*^ receiving gratuities 

167. The Supreme Court has found that it is not so vague or overbroad as to violate the due 
process clause of the fifth amendment. See Amett v. Kennedy, 416 U.S. 134, 159-60 (1974), reh'g 
denied, 417 U.S. 977 (1974). The Supreme Court refused to strike down the Lloyd-LaFollete Act 
in a suit brought by a former competitive service employee of the Office of Economic Opportunity. 
Kennedy had been removed by the agency for, among other charges, allegedly attempting to bribe 
a community representative. The Court noted that in framing the standard "to promote the efficiency 
of the service," Congress sought to "give myriad different federal employees performing widely 
disparate tasks a common standard of job protection." Id. at 159. 

168. 5 C.F.R. § 731.202 provides: 

(a) General. In determining whether its action will promote the efficiency of the 
service, OPM shall make its determination on the basis of: 

( 1 ) Whether the conduct of the individual may reasonably be expected to interfere 
with or prevent effective performance in the position applied for or employed in; or 

(2) Whether the conduct of the individual may reasonably be expected to interfere 
with or prevent effective performance by the employing agency of its duties and 

(b) Specific factors . Among the reasons which may be used in making a determi- 
nation under paragraph (a) of this section, any of the following reasons may be considered 
a basis for disqualification: 

( 1 ) Delinquency or misconduct in prior employment; 

(2) Criminal, dishonest, infamous or notoriously disgraceful conduct; 

(3) Intentional false statement or deception or fraud in examination or appointment; 

(4) Refusal to furnish testimony as required by § 5.3 of this chapter; 

(5) Habitual use of intoxicating beverages to excess; 

(6) Abuse of narcotics, drugs, or other controlled substances; 

(7) Reasonable doubt as to the loyalty of the person involved to the Government of 
the United States; or 

(8) Any statutory disqualification which makes the individual unfit for the service. 

(c) Additional considerations. In making its determination under paragraph (a) of 
this section, OPM shall consider the following additional factors to the extent that these 
factors are deemed pertinent to the individual case: 

( 1 ) The kind of position for which the person is applying or in which the person is 
employed, including its sensitivity; 

(2) The nature and seriousness of the conduct; 

(3) The circumstances surrounding the conduct; 

(4) The recency of the conduct; 

(5) The age of the applicant or appointee at the time of the conduct; 

(6) Contributing social or environmental conditions; 

(7) The absence or presence of rehabilitation or efforts toward rehabilitation. 

169. Chiriaco v. United States, 235 F. Supp. 850 (N.D. Ala. 1963), affd, 339 F.2d 588 (5th 
Cir. 1964). 


and accepting loans from a company with whom the employee had official 
dealings, '^^ behaving unbearably discourteously toward subordinates and co- 
workers,'^' failing to carry out a superior's orders, '^^ and exercising bad judg- 

Courts have required that there be a nexus between the unsatisfactory con- 
duct and the employee's job performance. Even immoral '^"^ or criminal '^^ con- 
duct does not support dismissal unless it is shown to be related to the employee's 
job performance. Many official abuses of power or commissions of torts might 
be considered severe enough to require the imposition of disciplinary sanctions 
to "promote the efficiency of the service." On the other hand, the official can 
always defend the action on the basis that the misconduct could not be "rea- 
sonably" expected to affect the job performance, or does not "reasonably" 
interfere with the agency's effective performance of its duties. 

An employee who is brought up on disciplinary charges to "promote the 
efficiency of the service" is entitled to advance written notice, an opportunity 
to answer the charges orally and in writing, representations by an attorney, and 
a written decision. '^^ Except in the case of a suspension for fourteen days or 
less, the employee has a right to appeal the agency's decision to the Merit 
Systems Protection Board. '^^ 

170. Monahan v. United States, 354 F.2d 306 (Ct. CI. 1965). 

171. Green V. Baughman, 243 F.2d 610 (D.C. Cir. \951), cert, denied, 355 V.S. 819(1957). 

172. Halsey v. Nitze, 390 F.2d 142 (4th Cir. 1968), cert, denied, 392 U.S. 939 (1968). 

173. Korman v. United States, 462 F.2d 1382 (Ct. CI. 1972). 

174. In Major v. Hampton, 413 F. Supp. 66 (E.D. La. 1976), the court overturned the 
dismissal of a married Internal Revenue Service agent who rented a "shack pad" in New Orleans. 
The court found no rational basis for concluding that the agent's conduct would discredit him or the 

175. Grebosz v. United States Civil Service Commission, 472 F. Supp. 1081 (S.D.N.Y. 

176. The specific applicable procedures are set out in 5 U.S.C. §§ 7503, 7513, and 7543. 
The procedures are somewhat less rigorous when only a suspension for fourteen days or less is being 
sought. The employee is not entitled to a hearing, but the agency may choose to provide one. 5 
U.S.C. §§ 7513(c), 7543(c). In Amett v. Kennedy, supra note 167, 156-58, the Supreme Court 
found that a hearing prior to the determination was not constitutionally required. However, the 
appeal mechanism does provide for a hearing after the administrative action has been taken. 

177. 5 U.S.C. §§ 7513(d), 7543(d). The Office of Personnel Management (0PM) regulations 
provide that 0PM may, for a period of one year after an appointment, "instruct an agency to remove 
an appointee when 0PM determines this action will promote the efficiency of the service." 5 C.F.R. 
§ 731.201. When 0PM acts to bring about the removal of an employee, it need not comply with 
all the procedures set out for removals initiated by the agencies themselves. 5 C.F.R. § 731.302(c); 
5 C.F.R. § 731.201. The legitimacy of this regulation is based on 5 U.S.C. §§ 3301 and 3302, 
which give the President (and thus 0PM) general powers to prescribe regulations so as to promote 
the efficiency of the service. Even after an employee has been working for more than one year, 
0PM may require the removal of an appointee who it finds made an intentional false statement or 
perpetrated a deception or fraud in the course of his or her examination or appointment. 5 C.F.R. 
§ 731.202. 


D. External Controls 

Executive officials are subject to sanctions from without as well as within 
the administrative system. In general, however, these sanctions depend on pub- 
licity, and are most commonly effective only against outrageous or notorious 
conduct. These sanctions are rarely exercised merely because of the harm suffered 
by a single individual. In addition, the sanctions usually apply only to current 
as opposed to former officials. While the latter continue to be liable for actual 
crimes they have committed, they are only subject to impeachment and the other 
sanctions described below if they have taken another public office. Former 
officials, of course, would continue to be vulnerable to adverse publicity. 

/ . Impeachment 

Under the Constitution "[t]he President, Vice President and all civil Officers 
of the United States, shall be removed from office on Impeachment for, and 
conviction of Treason, Bribery, or other high Crimes and Misdemeanors. "^^^ 
A vote of two thirds of the Senate is required to bring about such a removal. '^^ 
Conceivably therefore, if the President, Vice President, or other high ranking 
federal offical caused a serious harm to an individual, this official might be 
threatened with impeachment. But "high Crimes and Misdemeanors" may be 
interpreted quite narrowly, and more significantly it would be the extraordinary 
case that would overcome the institutional hurdles to impeachment. 

2. The Electorate 

Elections subject elected officials directly, and others indirectly, to incen- 
tives and sanctions for their performance. The effectiveness of their control hinges 
upon whether misconduct is publicized. The risk of losing an election because 
of voters' reactions to an individual case of wrongdoing by any elected official 
is minimal unless the incident involved criminal or particularly scandalous be- 
havior. Obviously, the impact on the election of the President and Vice President 
of the conduct of a single official is even less substantial. 

3. Congressional Control 

Congress may, at least theoretically, curb executive abuses of power in 
three ways. First, it may in some instances override or reverse administrative 
decisions it finds unacceptable. Second, it may seek to curb future abuses by 
enacting limiting legislation. Third, Congress may conduct oversight hearings. 
However, each of these sanctions is limited both by the fact that only a few very 
serious problems of a national magnitude can possibly be dealt with at the 
Congressional level and by separation of powers considerations. 

Many executive decisions are taken too quickly for Congress to have an 
opportunity to react. For example, if a law enforcement agent were to decide to 
conduct an illegal search. Congress could not possibly stop the search before it 

178. U.S. Const, art. II. § 4. 

179. U.S. Const, art. I, § 3. cl. 6. 


happened. Furthermore, since Congress cannot possibly be kept informed of the 
plans or acts of all government officials, it cannot serve as the regulatory body 
which insures that individual government officials do not abuse their discretion. 
Finally, there are constitutional limits on the extent to which Congress can 
intervene in executive branch affairs. '^^ 

This same concern arises with respect to congressional regulation of future 
agency activity. While Congress clearly has the power to abolish and create 
certain agencies, as well as to define their general purposes, it does not have 
unlimited authority to dictate precisely how executive branch affairs should be 
conducted.'^' Thus, while Congress may generally exert an influence by, for 
example, defining those agencies with which it is particularly displeased, or by 
passing limiting legislation,'^^ it is unrealistic to expect Congress to bear much 
of the burden of insuring that individual executive officials perform their jobs 

Like the impeachment and electorate sanctions, congressional oversight is 
highly dependent upon publicity. Unless the public can be made aware of and 
becomes outraged at executive abuses. Congress will probably not spend a great 
deal of its time attempting to discipline executive officials. 

4. Special Prosecutor 

High-level administrative officials '^^ are subject to investigation by a court- 
appointed Special Prosecutor if they are suspected of violating any federal crim- 
inal law other than a petty criminal offense. '^"^ Therefore, to the extent that 
federal criminal law overlaps with executive abuses of authority, the appointment 
of a Special Prosecutor may be regarded as a sanction against executive wrongs. 
This sanction is useful only in a very limited number of cases, and is unlikely 
to be employed unless public exposure of an executive official's alleged abuses 
have created a political demand for investigation. 

The law covering the appointment of Special Prosecutors requires that if 
the Attorney General receives information that high-level officials have com- 
mitted federal crimes, the Department of Justice is to conduct a preliminary 

180. See, e.g., Nixon v. Administrator of General Services, 433 U.S. 425 (1977) (holding 
that the Presidential Recordings and Materials Preservation Act did not violate the separation of 
powers principle, but noting that it is appropriate to examine the extent to which congressional 
legislation might prevent the executive from performing its function); Gregg v. Georgia, 428 U.S. 
153. 199 n. 50 (1976) (per Stewart, J., with Powell, J., and Stevens, J., concurring and White, 
J., Burger, J.. Rehnquist C.J., and Blackmun, J., concurring in part with the judgment) (it would 
be unconstitutional to prohibit the President from granting a reprieve from a death sentence as matter 
of executive clemency). 

181. Gregg, supra note 180. 

182. See, e.g., the War Powers Resolution, 50 U.S.C. §§ 1541-48, requiring the President 
to consult Congress before taking certain military actions. 

183. Special prosecutors are to be appointed to investigate the President and the Vice President, 
Cabinet officials, high-level Executive Office employees, and certain campaign officials. 28 U.S.C. 
§ 591. 

184. 28 U.S.C. § 591. 


investigation. Based on this investigation the Attorney General may recommend 
that no Special Prosecutor be appointed.'^** If the Attorney General does not so 
recommend, he or she shall apply for the appointment of a Special Prosecutor. '^^ 
Special statutory provisions cover the authority for removal of a Special Pros- 
ecutor,'^^ who may bring suits based on his investigations, and submission to 
Congress of reports on his activities. 

Therefore, while Special Prosecutors can in certain situations conduct rel- 
atively independent investigations from within the administration, this institution 
is unlikely to prove very effective as a general means of dealing with wrongs 
committed by executive officials. 

5. Public Pressure 

Public pressure plays an important role in correcting and deterring improper 
activities of government officials. The sanctions such as impeachment, non- 
reelection, and the appointment of a Special Prosecutor, are usually imposed 
only when sufficient public pressure is brought to bear. However, such pressure 
can be brought to bear only if a wrongful act is accompanied by sufficient 
publicity and then only if the public is able to make its discontent known. 
Therefore, it is unrealistic to presume that substantial public pressure could 
generally be aroused with respect to the activities of most government officials. 

The public lacks the investigative and informational capacity to determine 
whether each of the thousands of federal executive officials are carrying out their 
jobs properly. Thus, unless a low-level official engaged in either extreme or 
repeatedly wrongful actions, the public would probably not become aware that 
a wrongful act had occurred. Moreover, it will generally prove difficult to 
organize public pressure against an official who has not harmed a substantial 
number of persons. 

The Role of Administrative and External Controls in the Civil Sanction System 
Existing internal administrative and independent external controls provide 
means by which government officials may be sanctioned for their wrongdoing. 
However, the current federal system for administrative disciplinary controls was 
not developed primarily to regulate conduct that directly harms the public, and 
for this reason cannot fully serve all the goals related to official liability and 
immunity rules such as compensation of victims and deterrence. 

Furthermore, former officials are not subject to internal disciplinary pro- 
cedures. Nor are they subject to impeachment or removal from office. '^^ Other 

185. 28 U.S. C. § 592(b)(1). 

186. Under 28 U.S.C. § 49, a panel of judges from the Circuit Court of Appeals for the 
District of Columbia is responsible for appointing the Special Prosecutor. 

187. 28 U.S.C. S§ 594-96. 

188. However, a government official who has left one government position but taken on a 
new one may be subject to certain disciplinary sanctions for actions taken in the old job. 



officials, those appointed or elected, are subject only to sanctions rarely employed 
as a result of harms suffered by individual members of the public. 

Nevertheless, while the existing system of disciplinary and other remedies 
does not satisfy all the goals of a system dealing with governmental wrongdoing, 
it is a good first step toward such a system. Most importantly, it provides an 
established mechanism with sufficient resources and expertise to investigate the 
acts of numerous individual government officials. Thus, in developing an alter- 
native method of dealing with governmental wrongdoing one should build on 
this existing system, but adjust it as well, so that it covers high ranking officials 
as well as civil servants and so that it explicitly applies to misconduct that results 
in harm to individual members of the public. 


IV. Analysis: Changes at the Federal Level 

There is almost universal dissatisfaction with the current system of civil 
sanctions for violations of constitutional rights by federal officials. This dissat- 
isfaction stems, in part, from the recognition that private damage actions against 
individual government officials do not afford the victims of official misconduct 
"a financially responsible defendant."' Of equal concern to many is the per- 
ception that the threat of personal liability — or, at a minimum, the risk of suit — 
deters not only improper conduct by government officials, but proper conduct 
as well.^ Adding to these pragmatic concerns is the conviction that the govern- 
ment ought to be liable for the wrongs committed in its name, rather than impose 
the loss upon an individual official ''who is more anvil than hammer. ""^ Finally, 
some critics of the current system cite the supposed cost to the government of 
retaining private counsel to defend individual government officials in Bivens 
suits. "^ 

1 . Federal Tort Claims Act: Hearings Before the Subcommittee on Administrative Law and 
Governmental Relations of the House Committee on the Judiciary, 95th Cong., 2d Sess. 4 (1978) 
(statement of Attorney General Griffin B. Bell) (hereinafter cited as 1978 House Hearings); see Tort 
Claims: Hearings on H.R. 24, H.R. 3060, and H.R. 3795 Before the Subcommittee on Administrative 
Law and Governmental Relations of the House Committee on the Judiciary, 97th Cong., 1st Sess. 
27 (1982) (statement of Edward C. Schmults, Deputy Attorney General) (hereinafter cited as 1982 
House Hearings). 

2. See 1978 House Hearings, supra note 1, at 4 (statement of Attorney General Bell); 1982 
House Hearings, supra note 1, at 27 (statement of Deputy Attorney General Schmults). 

3. Schuck, Suing Our Servants: The Court, Congress, and the Liability of Public Officials 
for Damages, 1980 Sup. Ct. Rev. 281, 347, reprinted in Federal Tort Claims Act: Hearing on S. 
1775 Before the Subcommittee on Agency Administration of the Senate Committee on the Judiciary 
(Pt.2), 97th Cong., 2d Sess. 35 (1982) (hereinafter cited as 1982 Senate Hearing). 

4. Bivens v. Six Unknown Named Agents, 403 U.S. 388 (1971). See, e.g.. Amendments to 
the Federal Tort Claims Act: Joint Hearing on S. 21 17 Before the Subcommittee on Citizens and 
Shareholders Rights and Remedies and the Subcommittee on Administrative Practice and Procedure 
of the Senate Committee on the Judiciary (Pt.l), 95th Cong.. 2d Sess. 7 (1978) (statement of 
Attorney General Bell) (hereinafter cited as 1978 Senate Hearing). Employment of such private 
counsel is often militated by ethical or policy considerations. See 1982 House Hearings, supra note 
1, at 241-42 (statement of Stanley S. Harris, U.S. Attorney for the District of Columbia); Amend- 
ments of the Federal Tort Claims Act: Hearings Before the Subcommittee on Administrative Law 
and Governmental Relations of the House Committee on the Judiciary, 96th Cong., 1st Sess. 3 
(1979) (statement of Deputy Attorney General Benjamin R. Civiletti) (hereinafter cited as 1979 
House Hearings). It has been estimated that the Justice Department has spent over $2 million for 
such private counsel since 1976. 1982 House Hearings, supra note 1, at 24 (testimony of Deputy 
Attorney General Schmults); Bell, Proposed Amendments to the Federal Tort Claims Act, 16 Harv. 
J. on Legis. 1, 9 n. 28 (1979). 

The actual economy to the federal government resulting from an enterprise liability system is 
uncertain at best. See 1979 House Hearings, supra, at 12 (Department of Justice Memorandum, 
Cost estimate for the proposed amendments to Federal Tort Claims Act (May 7, 1979)). Insofar as 
substituting the United States as defendant would afford plaintiffs "a financially responsible defen- 
dant," it is far from obvious that adopting an enterprise liability system would save the government 
any money at all, except to the extent that legislation establishing such liability would definitively 
permit the government to settle rather than pay to litigate cases involving alleged employee mis- 
conduct. See 1982 House Hearings, supra, note 1 at 248 (statement of Royce C. Lamberth, Chief, 


Animated by these and other subsidiary concerns, critics of the current 
system of civil sanctions have sought for the past ten years its modification or 
replacement — at least at the federal level — by a system of "enterprise liability" 
in which the government, rather than the individual government official, would 
be liable in damages for constitutional injuries caused by official misconduct.^ 
Substituting the government for the individual official as defendant in civil 
damage actions would promote the goal of compensating victims of such mis- 
conduct, while the goals of deterrence and employee accountability would be 
served by a system of administrative rather than judicial sanctions.^ 

Despite wide support for the broad goals of enterprise liability, legislative 
efforts to establish such a system have so far foundered on disagreements over 
appropriate substitute mechanisms of accountability. Failure to reach agreement 

Civil Division, Office of the U.S. Attorney for the District of Columbia) (arguing that substitution 
of United States as defendant would facilitate settlement and reduce litigation). 

It thus appears that the major gain from the standpoint of government policy would be to assure 
the government's complete control over the defense of constitutional tort actions. Cf. id. at 2-4 
(statement of U.S. Attorney Harris); 1979 House Hearings, supra, at 3-4 (statement of Deputy 
Attorney General Civiletti). 

5. See H.R. 10439. 93rd Cong., 1st Sess. (1973); S. 2558, 93rd Cong., 1st Sess. (1973); 
H.R. 12715, 93rd Cong., 2d Sess. (1974); H.R. 9219, 95th Cong., 1st Sess. (1977); H.R. 9437, 
95th Cong., 1st Sess. (1977); S. 21 17, 95th Cong., 1st Sess. (1977); S. 2868, 95th Cong., 2d Sess. 
(1978); S. 3314, 95th Cong., 2d Sess. (1978); H.R. 193, 96th Cong., 1st Sess. (1979); H.R. 2659, 
96th Cong., 1st Sess. (1979); S. 695, 96th Cong., 1st Sess. (1979); H.R. 24, 97th Cong., 1st Sess. 
(1981); H.R. 1696, 97th Cong., 1st Sess. (1981); S. 1775, 97th Cong., 1st Sess. (1981); H.R. 
6359, 97th Cong., 2d Sess. (1982); H.R. 7034, 97th Cong., 2d Sess. (1982); H.R. 595, 98th Cong., 
1st Sess. (1983). Copies of the foregoing bills are reproduced in Appendix A-1 hereto, together 
with a summary and chronology of each. 

In 1974, Congress amended the Federal Tort Claims Act, 42 U.S.C. §§ 1346(b), 2671-2680 
("FTCA"), to permit suits against the government for assault, battery, false arrest, abuse of process, 
and malicious prosecution by federal investigative or law enforcement officers. Act of Mar. 16, 
1974, Pub. L. No. 93-253, § 2, 88 Stat. 50 (codified at 28 U.S.C. § 2680(h)). Such torts are 
arguably coterminous with claims arising under the due process clause of the fifth amendment and 
other provisions of the Bill of Rights. See infra Part IV-B(2)(g). Having enacted Pub. L. No. 93- 
253, the 93rd Congress failed to act on alternative legislation proposed by the Nixon Administration 
that would have broadened the FTCA to cover all claims of unconstitutional torts involving any 
government official. Federal Tort Claims Amendment: Hearings on H.R. 10439 Before the Sub- 
committee on Claims and Governmental Relations of the House Committee on the Judiciary, 93rd 
Cong., 2d Sess. 14-15 (1974) (statement of Irving Jaffe, Acting Assistant Attorney General, Civil 
Division, Department of Justice). See also Intelligence Activities and the Rights of Americans: Final 
Report of the Senate Select Committee to Study Governmental Operations with Respect to Intelligence 
Activities, S. Rep. No. 755, 94th Cong., 2d Sess. 336-38 (1976) (recommending statutorily-created 
federal cause of action against the government and individual federal intelligence officer or agent 
for actual or threatened injury caused by a violation of the Constitution committed under color of 

6. See infra Part IV-B(2)(b). Calls for a system in which government officials would be 
indemnified either by the government or under an insurance program have failed to gain currency, 
presumably because such as system could not eliminate the debilitating effect on employee morale 
and decisionmaking vigor said to result from the risk of suit. In addition, insurers are reportedly 
unwilling to cover at least some federal officials. See 1978 House Hearings, supra note 1, at 31 
(statement of John S. McNemey, National President, Federal Criminal Investigators Association). 


Upon such mechanisms reflects the fact that assuring redress and deterrence, and 
fostering vigorous decisionmaking by federal officials, are often seen as incom- 
patible goals. ^ Thus, although legislative proposals to establish some form of 
enterprise liability at the federal level have been sounded since shortly after 
Bivens itself was decided,^ none has yet been adopted.^ 

Legislation pending in the 97th Congress would have amended the Federal 
Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680, to provide an exclusive 
remedy against the United States for constitutional torts committed by govern- 
ment officials while acting within the scope of their office or employment. The 
following section analyzes that proposed legislation — H.R. 7034 (reintroduced 
in the 98th Congress as H.R. 595) and S. 1775 — and considers a number of 
issues raised by those and other bills that illustrate the principal problems involved 
in establishing a system of enterprise liability at the federal level. Analysis of 
these bills may facilitate consideration of any legislation that may be introduced 
on this subject in the 98th Congress. 

A. Summary of Issues Raised by Proposals to Establish Federal 

Enterprise Liability 

/ . Proposed Legislation 

(a) H.R. 7034.^^ 

H.R. 7034 would have amended the Federal Tort Claims Act ("FTCA") 
to provide that the United States would be exclusively liable with respect to a 
claim arising under the Constitution of the United States for torts committed by 

7. See 1978 Senate Hearing, supra note 4, at 8 (statement of Attorney General Bell). The 
current Administration has taken the position that: 

The problem here is not really Government employees acting in an unlawful fashion . . . 
[but rather] the overhang of litigation and the burden on Government employees of having 
... to be worried about their own cases. 

Federal Tort Claims Act: Hearings on S. 1775 Before the Subcommittee on Agency Administration 
of the Senate Committee on the Judiciary (Pt.l), 97th Cong., 1st Sess. 7-8 (1981) (statement of 
Deputy Attorney General Schmults) (hereinafter cited as 1981 Senate Hearings). The previous 
Administration, by contrast, took the view that the problem of deterrence was sufficiently serious 
that a new employee discipline proceeding was warranted "to replace the sanction against employee 
misconduct presumably lost by immunizing employees from civil liability." 1979 House Hearings, 
supra note 4, at 4 (statement of Deputy Attorney General Civiletti). 

8. See supra note 4. 

9. Of the 17 bills reproduced in Appendix A-1, only three— H.R. 9219, S. 3314, and H.R. 
7034 (in lieu of H.R. 24) — were even forwarded to full committee by the responsible subcommittee. 
None was ever reported out of committee. 

10. H.R. 7034 was pending in the House Judiciary Committee at the end of the 97th Congress, 
having been approved and forwarded by the Subcommittee on Administrative Law and Governmental 
Relations. H.R. 7034 was reintroduced in the 98th Congress as H.R. 595. 129 Cong. Rec. H94 
(daily ed. Jan. 6, 1983). 



government officials acting within the scope of their office or employment. '^ 
With respect to any constitutional tort claim, the United States would have been 
made liable for the greater of either actual damages or liquidated damages, which 
would have been assessed in an amount which was the greater of either (A) 
$2,000 or (B) in the case of a continuing violation, $200 per day for each 
violation.'^ If it were established that the defendant official acted with malicious 
intent or reckless disregard for the plaintiff's constitutional rights, "additional" 
damages of up to $100,000 would have been authorized.'^ A successful claimant 
would also have been entitled to receive reasonable attorney fees and all other 
litigation costs reasonably incurred, inlcuding attorney fees or costs attributable 
to processing an administrative claim for money damages based on the alleged 
constitutional tort.'"^ 

Under H.R. 7034, a suit against an individual government official for an 
alleged constitutional tort would have become a suit against the United States 
"[u]pon certification by the Attorney General that the defendant employee was 
acting within the scope of his office or employment at the time of the incident 
out of which the suit arose. . . ."'^ The United States would have been free to 

n. H.R. 7034, § 202 (proposed 28 U.S.C. § 2699(b)). 

Both H.R. 7034 and S. 1775 would have provided an exclusive remedy against the United 
States not only with respect to constitutional torts, but also generally in suits based upon acts or 
omissions of United States officials committed within the scope of office or employment. See H.R. 
7034, § 101 (proposed 28 U.S.C. § 2679(b)); S. 1775, § 5(a) (same). Currently, a plaintiff may, 
as a rule, sue both the official and the government for the same conduct, stating a constitutional 
tort claim against the former under Bivens, and a non-constitutional tort claim against the latter 
under the FTC A. See Carlson v. Green, 100 S. Ct. 1468, 1472 (1980) (construing 28 U.S.C. § 
2680(h)). Only in a limited set of cases has Congress explicitly made suit against the government 
under the FTCA an exclusive remedy. See 28 U.S.C. § 2679(b) (operation of motor vehicles by 
federal employees); 38 U.S.C. § 4116(a), 42 U.S.C. §§ 233(a), 2458(a) (malpractice by certain 
government health personnel). The statutory enshrinement of a distinction between constitutional 
and non-constitutional torts has been criticized as unsound and unnecessary. See, e.g., 1979 House 
Hearings, supra note 4, at 23 (statement of George A. Bermann, Professor of Law, Columbia 
University Law School). 

12. H.R. 7034, § 202 (proposed 28 U.S.C. § 2693). Proposals to replace the Bivens remedy 
have typically provided for liquidated damages to assure successful plaintiffs monetary recovery 
where actual damages are small or nonexistent, on the premise that proof of concrete injury "is 
often impossible in the case of nonviolent conduct involving violations of constitutional rights which 
are of an intangible nature." 1978 Senate Hearing, supra note 4, at 8 (statement of Attorney General 
Bell). A liquidated damages provision by definition establishes both a floor and a ceiling on the 
government's liability in cases where it is appropriate to award such damages. See infra note 44. 

13. H.R. 7034, § 202 (proposed 28 U.S.C. § 2693). Notwithstanding the provision for such 
"additional" damages, H.R. 7034 purported to preclude "punitive damages." Id. ([t]he United 
States . . . shall not be liable ... for punitive damages"). 

14. H.R. 7034. § 202 (proposed 28 U.S.C. § 2698(a)). Under H.R. 7034, an action could 
not have been instituted against the United States unless the claim had first been presented to, and 
denied by, the appropriate federal agency. Id. (proposed 28 U.S.C. § 2694(a)). A similar conditon 
applies to suits against the United States under the FTCA. See 28 U.S.C. § 2675. 

15. H.R. 7034, § 202 (proposed 28 U.S.C. § 2698(d)(1)). Certification by the Attorney 
General would have triggered removal to federal court of a suit brought in a state court. Id. § 202 
(proposed 28 U.S.C. § 2698(d)(2)). The Attorney General's certification that the defendant official 
was acting within the scope of his office or employment at the time of the incident would have been 


assert as a defense to a constitutional tort claim "the absolute or qualified 
immunity of the employee whose act gave rise to the claim, or his reasonable 
good faith belief in the lawfulness of his conduct."'^ 

Finally, H.R. 7034 provided that, where a constitutional tort action resulted 
in a judgment against the United States or an award, compromise, or settlement 
paid by the United States, "the Attorney General shall forward the matter for 
such further administrative investigation or disciplinary action as may be appro- 
priate to the head of the department or agency which employed the employee 
at the time of the employee's alleged act or omission giving rise to the claim. "'^ 
H.R. 7034 was drafted so as to preserve the right to a jury trial in constitutional 
tort cases, a right now provided by the Bivens remedy.'^ 

(b) S. 1775'"^ 

S. 1775 was substantially similar to H.R. 7034, with three significant 
exceptions. First, S. 1775 would have provided that the individual official and 
the United States would be jointly liable for punitive damages of up to $50,000 
for the official's willful and malicious violation of constitutional rights. ^^ Second, 
no provision was made under S. 1775 for any award to a prevailing plaintiff of 
attorney fees or other litigation costs. And, third, S. 1775 would not have 
provided for jury trial in all constitutional tort cases, but only in cases either 
pending on the date of the enactment of the bill, or based on claims that had 
arisen as of that date.*^' 

2. Assessment of Issues Raised by H.R. 7034 and S. 1775 

As noted above, there is nearly unanimous support for the proposition that 
the United States, and not individual federal officials, should bear the ultimate 

''binding and conclusive [only] on the United States and the defendant employee," id. § 202 
(proposed 28 U.S.C. § 2698(d)(3)), and not on the party bringing the action. 

If a plaintiff successfully challenged the Attorney General's certification upon removal, the 
district court would have been required to remand the cause to the state court, inasmuch as the 
Attorney General's certification was the predicate for removal. Id. If the Attorney General did not 
make such a certification, the defendant employee would have been allowed to request the federal 
district court — but apparently not the state court — in which the suit has been filed to certify that 
the defendant was acting within the scope of his office or employment at the time of the alleged 
injury. Id. 

16. H.R. 7034. § 202 (proposed 28 U.S.C. § 2693). 

17. H.R. 7034, § 202 (proposed 28 U.S.C. § 2700). 

18. Id. § 303(a). Significantly, the power to award "additional" damages for malicious 
constitutional torts was vested with the court, not the jury. See supra note 13. 

19. S. 1775. 97th Cong.. 1st Sess., 127 Cong. Rec. S12, 152-53 (daily ed. Oct. 26, 1981). 
S. 1775 was pending in the Subcommittee on Agency Administration of the Senate Judiciary 
Committee at the end of the 98th Congress. Senator Grassley. Chairman of the Subcommittee, was 
expected to introduce new legislation in the 98th Congress. 

20. S. 1775, § 3 (proposed 28 U.S.C. § 2674(b)(3)) (amendment of Sen. Specter, adopted 
(Sept. 23, 1982)). The United States would have been entitled to recover the amount paid under 
this subsection from the official. Id. Under S. 1775, the maximum amounts allowed for actual and 
liquidated damages were also lower than those set by H.R. 7034. Id. See infra note 44. 

21. S. 1775, § 10(b)(1). 


financial responsibility for constitutional violations. ^-^ But differences have inev- 
itably arisen over central elements of any new system of exclusive enterprise 
liability, and both H.R. 7034 and S. 1775 aptly illustrate the dilemmas facing 
efforts to establish such a system. 

(a) Retention of the "Good Faith" Defense. 

The most hotly disputed issue in establishing an exclusive system of en- 
terprise liability is whether the government should be allowed, as H.R. 7034 
and S. 1775 provided, to assert as a defense the official's ''absolute or qualified 
immunity ... or his reasonable good faith belief in the lawfulness of his con- 
duct. "^^ 

For obvious reasons, the availability of the "good-faith" defense would — 
especially after Harlow v. Fitzgerald — significantly limit the extent of the gov- 

22. But see 1982 House Hearings, supra note 1 , at 43 1 . 435-40 (statement of Thomas Devine 
on behalf of the Government Accountability Project ("GAP") of the Institute for Policy Studies). 
Disputing the premise that the threat of personal liability "functionally paralyzes any conscientious 
federal employee," id. at 435, and that judgment-proof government officials thwart the goal of 
victim compensation, id. at 438-39, GAP has proposed a modified version of H.R. 7034 that would 
supplement, but not supplant, the Bivens-type remedy. Id. at 451. Similarly, Professor Neubome 
has suggested that the Federal Tort Claims Act simply be amended to permit Bivens defendants to 
implead the United States, establishing "a self-executing mechanism [that] would allocate the 
economic loss caused by a constitutional violation to the proper party." 1981 Senate Hearings, 
supra note 7, at 29 (statement of Professor Burt Neubome on behalf of ACLU). 

These proposals for an alternative to an exclusive system of enterprise liability echo the joint 
government/employee liability recommended by the Senate Select Committee, S. Rep. No. 755, 
supra note 5, at 336-38 (1974), and the Senate Committee on Government, S. Rep. No. 588, 93rd 
Cong., 1st Sess. 2-4, reprinted in 1974 U.S. Code Cong. & Ad. News 2789 (analyzing proposed 
amendment to H.R. 8245) — ultimately enacted into law as Pub. L. No. 93-253, supra note 5 — to 
provide, among other things, a remedy against the United States for certain intentional non-consti- 
tutional torts of its investigative and law enforcement officers). Attorney General Bell did not favor 
allowing a plaintiff to choose between suing the employee and suing the government. 1978 Senate 
Hearing, supra note 4, at 1 19 (Letter from Justice Department to Sen. Abourezk (Mar. 13, 1977)). 

23. H.R. 7034, § 202 (proposed 28 U.S.C. § 2693): S. 1775, § 5(b) (proposed 28 U.S.C. 
§ 2679(d)(2)). Cf. Norton v. United States, 581 F.2d 390, 394-95 (4th Cir.), cert, denied, 439 
U.S. 1003 (1978) (holding that Congress intended to allow the government to assert an official's 
good faith in suits brought under 28 U.S.C. § 2680(h)). The Supreme Court appears to define 
qualified immunity as the right to invoke the good-faith defense. Harlow v. Fitzgerald, 102 S. Ct. 
2727, 2737 (1982); see infra note 24. For convenience, the issue of whether to allow the government 
to assert either the absolute or qualified immunity of its officials is discussed herein under the rubric 
of whether to allow the government to assert the good-faith defense. 

Several earlier bills, in contrast to H.R. 7034 and S. 1775, would have precluded the United 
States from asserting— except in cases involving a Member of Congress, a judge, a prosecutor, or 
a person performing analogous functions — the absolute or qualified immunity of the official whose 
conduct was at issue, or the official's reasonable good-faith belief in the lawfulness of his conduct. 
See S. 3314, supra note 5, § 3(b); H.R. 2659, supra note 5, § (3); S. 695, supra note 5, § 3; H.R. 
24, supra note 5, § 3. This approach was supported by the previous Administration. See infra note 
26. Under S. 2117, supra note 5, § 3, and under H.R. 9219, supra note 5, § 3, a federal officer's 
or employee's "good faith reliance on a court order or legislative authorization" would have 
constituted a complete defense in a constitutional tort suit against the United States, but the United 
States would nevertheless have been required to compensate the person whose constitutional rights 
had been violated. 


emment's liability for the constitutional torts of its officials, and thus the avail- 
ability of damage awards to the victims of such torts to redress violations of 
constitutional rights. ^"^ 

Moreover, allowing the government to assert the good-faith defense would 
also limit the operation of any disciplinary mechanism that would be triggered 
by a determination of liability or settlement by the govemment;^^ such mechanism 
would — at least in the view of the current Administration — be activated only 
when the official had intentionally violated the plaintiff's constitutional rights. ^^ 

24. 102 S. Ct. 2727, 2736-39 ( 1982). Defining •'[qjualified or 'good faith' immunity," the 
Supreme Court held that "government officials performing discretionary functions generally are 
shielded from liability for civil damages insofar as their conduct does not violate clearly established 
statutory or constitutional rights of which a reasonable person would have known." Id. at 2738. 
This "[r]eliance on the objective reasonableness of an official's conduct, as measured by reference 
to clearly established law, ' ' id. at 2739, supersedes reliance both on objective factors and on subjective 
factors {i.e., "permissible intentions") in determining whether the official acted in good faith. Id. 
at 2737. Whether the adjustment of the good-faith defense to rely wholly on objective factors will 
actually reduce the extent of liability for official misconduct, or will simply "permit the resolution 
of many insubstantial claims on summary judgment," id. at 2739, is highly doubtful, for although 
Harlow may have somewhat simplified trial of the good faith issue with respect to the question of 
unclear applicable law. Other subjective elements certainly remain to be asserted or denied in 
establishing the official's state of mind. See, e.g., Briggs v. Goodwin, No. 80-2269 (D.C. Cir. Jan. 
11, 1983); McSurely v. McClellan, No. 82-2369 (D.C. Cir. Dec. 10, 1982); Dale v. Bartels, No. 
74-Civ. 1382-CLB (S.D. N.Y., Dec. 17. 1982); cf. Sampson v. King, 693 F.2d 566 (5th Cir. 
1982); Saldana v. Garza, 684 F. 2d 1 159 (5th Cir. 1982); United States v. Irving, 684 F.2d 494 (7th 
Cir. 1982); Standridge v. City of Seaside, 545 F. Supp. 1195 (N.D. Cal. 1982); Thompson v. 
Pennsylvania Parole Board Member Jefferson, 544 F. Supp. 173 (E.D. Penn.1982). 

25. See H.R. 7034, S 202 (proposed 28 U.S.C. 8 2700); S. 1775, § 5 (proposed 28 U.S.C. 
§ 2679(f)). 

26. To be sure, even if the government were permitted to assert the good-faith defense, the 
agency that employed the official whose conduct was challenged would theoretically be free to 
initiate its own internal investigation and disciplinary proceedings under existing law at any time, 
irrespective of the pendency of the lawsuit. But the current Administration has strongly implied that 
a determination of liability or money settlement by the United States would be a necessary predicate 
for disciplinary action against the defendant official. See 1 98 1 Senate Hearings, supra note 7, at 9 
(testimony of Deputy Attorney General Schmults) (arguing against waiver of good faith on ground 
that disciplinary proceedings would be triggered by determination of liability, and that where the 
official has acted in good faith "there should be no disciplinary proceedings"). 

As initially drafted (in the form of H.R. 24), H.R. 7034 was criticized by some for eliminating 
the "good-faith" defense in constitutional tort suits. Se^ 1982 House Hearings, supra note 1, at 
341 (statement of J. Bryan Hyland, President, Association of Federal Investigators); id. at 343-45 
(statement of Ira M. Lechner, Legislative Counsel, National Association of Police Organizations, 
Inc.). H.R. 24's original "good faith" waiver was specifically endorsed by, among others, the 
Federal Executive and Professional Association ("FEPA"), see id. at 328 (statement of Richard K. 
Pelz, President of FEPA), and waiver of the defense under earlier legislative proposals was supported 
by the FBI, see 1979 House Hearings, supra note 4, at 80 (statment of William H. Webster, Director, 
FBI) and by the Carter Administration, see 1978 House Hearings, supra note 1, at 14 (statement 
of Irving Jaffee, Deputy Assistant Attorney General, Civil Division, Department of Justice), as well 
as by the American Bar Association, 7979 House Hearings, supra note 4, at 152 (statement of B. 
James George, Jr., Immediate Past Chairperson, Criminal Justice Section, American Bar Associa- 


Because permitting the government to assert the official's good faith would 
thus undercut the goals of assuring victim compensation and reinforcing official 
accountability, those who would preserve the good-faith defense for the gov- 
ernment in constitutional tort suits bear a heavy burden of justification for their 
position. The argument advanced in support of retaining the good-faith defense 
for the government does not meet that burden. 

At the outset, it should be understood that under established tort doctrine 
applicable to every analogous situation of private and public vicarious liability, 
the United States would not be permitted to invoke the "good-faith" immunity 
of its officials in constitutional tort suits. At common law, the majority rule in 
the private employment context is that an employer sued for the tort of an 
employee under the doctrine of respondeat superior may not invoke the em- 
ployee's immunities. -^^ Moreover, the majority common-law rule in the public 
employment context at the state and local level is that a government agency sued 
for the tort of a public official under the doctrine of respondeat superior may 
not invoke the official's immunities. ^^ Indeed, in FTC A suits, virtually every 
federal court that has considered the issue has concluded that (a) the doctrine of 
respondeat superior does not permit the United States to invoke the immunities 
of its officials, ^^ and (b) Congress did not intend for the United States to be able 
to invoke the immunities of its officials in suits for intentional ' torts brought 
under 28 U.S.C. § 2680(h). ^° 

27. See, e.g., Schubert v. August Schubert Wagon Co., 249 N.Y. 253, 164 N.E. 42 (1928); 
Hallmark Insurance Co. v. Crary Enterprises, Inc., 72 Wis. 2d 472, 475, 241 N.W.2d 171, 173 
(1976); Restatement (Second) of Agency § 217 Comment b (1958); 2 F. Harper & F. James, The 
Law of Torts, § 26.17. at 1427 n.6 (1956); W. Seavey, Handbook on the Law of Agency, § 93, 
at 167 (1964). 

28. See, e.g., James v. Prince George's County, 288 Md. 315, 418 A. 2d 1 173, 1 182 (1980); 
Maynard v. City of Madison, 101 Wis. 2d 273, 304 N.W.2d 163, 169 (1981); Mutnan v. City of 
Monongahela, 45 Pa. Commw. 23, 406 A. 2d 81 1, 813-14 (1979). 

29. See Downs v. United States, 382 F. Supp. 713, 749-51 (M.D. Tenn. 1974), ajfd in 
pertinent part, 522 F.2d 990, 998 (6th Cir. 1975); United States v. Massachusetts Bonding & 
Insurance Co., 227 F.2d 385, 387 (1st Cir. 1956), rev'd on other grounds, 352 U.S. 128 (1958); 
Ray V. United States, 228 F.2d 574, 585 n.lO, 586 n.l2 (5th Cir. 1955) (Brown, J., dissenting); 
United States v. Trubow, 214 F.2d 192, 196 (9th Cir. 1954); Jackson v. United States, 196 F.2d 
725, 726 (3d Cir. 1952); United States v. Hull, 195 F.2d 64, 68 (1st Cir. 1952); The only decision 
contra appears to be Brooks v. United States, 152 F. Supp. 535, 537 (S.D.N.Y. 1957) (alternative 
holding), relying on the premise that common-law respondeat superior entitles an employer to invoke 
his employee's immunities. 

30. Four of the five federal courts to have considered the issue have reached this conclusion. 
The Fourth Circuit, declining to rely on the common-law doctrine of respondeat superior, has held 
that Congress intended to permit the United States to plead the qualified immunity of its officials 
in suits under 28 U.S.C. § 2680(h). Norton v. United States, 581 F.2d 390, 393-96 (4th Cir. 1978), 
rev'g All F. Supp. 138, 146-52 (E.D. Va. 1977). In addition to the district court in Norton, the 
three other federal courts that have considered the issue have reached the opposite conclusion. See 
Picariello v. Fenton, 491 F. Supp. 1026, 1040-42, (M.D. Pa. 1980); Townsend v. Carmel, 494 F. 
Supp. 30, 36-37 (D.D.C. 1980); Crain v. Krehbiel, 443 F. Supp. 202, 215-17 (N.D. Cal. 1978). 
Cf. Downs V. United States, 522 F.2d 990, 998 (6th Cir. 1975). See also Comment, Sovereign 
Immunity, 41 Geo. Wash. L. Rev. 651 (1979) (criticizing Fourth Circuit's Norton decision). 


Allowing the United States to plead the "good-faith" immunity of federal 
officials in constitutional tort suits would not appear to place the United States 
on a par with employers who face liability under respondeat superior?^ On the 
contrary, the government would have an advantage enjoyed neither by private 
nor public employers at common law, nor by the United States itself in non- 
constitutional tort suits. 

The principal argument for retaining the good-faith defense asserts, in es- 
sence, that no genuine wrongdoing has occurred where the official has acted in 
good faith, and that a judgment against the government in such cases would 
cause an unwarranted imputation of wrongdoing on the defendant official's part.''^ 

This argument simply cannot withstand scrutiny. The good-faith defense is 
made available to the individual government official not because the official's 
good faith somehow negates the fact of constitutional injury, or because a de- 
termination of liability would unfairly stigmatize him. Instead, "underlying the 
qualified immunity which public officials enjoy for actions taken in good faith 
is the fear that exposure to personal liability would deter them from acting at 
all."^-^ Thus, although a government official's good faith may be relevant to 
whether a party whose constitutional rights have been violated should be entitled 
to compensation from the official, the official's good faith cannot be relevant 
to whether the victim is entitled to compensation from the government. ^"^ 

31. FBI Director Webster, in fact, has suggested that the apt analogy in this context might 
be not to respondeat superior doctrine at all, but to decisional law denying insurance companies 
the right to assert the immunities of those whom they have insured. 1978 House Hearings, supra 
note 1, at 97. 

32. See 1982 House Hearings, supra note 1, at 343-44 (statement of Ira M. Lechner. Leg- 
islative Council, Nat'l Ass'n of Police Organizations, Inc.); id. at 334 (statement of Ordway P. 
Burden. President. Law Enforcement Assistance Foundation); id. at 138 (statement of Joseph A. 
Morris. General Counsel. Office of Personnel Management). 

Viewed in this light, the good-faith defense is not an excuse for misconduct, but rather a denial 
that any injury has occurred. Thus. Deputy Attorney General Schmults has frankly stated the 
Administration's view that the good-faith defense: 

[R]eally goes to the merits of the plaintiffs claim by testing the action of an employee 
against the standard of reasonableness and good faith. . . . [a]nd certainly no employee 
wants to be found 'guilty.' if you will, of unconstitutional acts and suffer the resulting 
stigma. So that, even where the United States and not the employee would be the defendant, 
our view is that employees would be discouraged from acting in uncertain areas where 
they might subject the Government to financial liability. 

1981 Senate Hearings, supra note 7. at 9. 

33. Carlson v. Green, 100 S. Ct. at 1473 n.7, citing Butz v. Economou. 438 U.S. 478. 497 
(1978); Scheuer v. Rhodes. 416 U.S. 232, 240 (1974). See Owen v. City of Independence. 445 
U.S. 612 (1980). As FBI Director William H. Webster has stated, the good-faith defense "is for 
the individual protection of the agent." 1978 House Hearings, supra note 1, at 97 (testimony). 

34. See 1979 House Hearings, supra note 4. at 126-27 (statement of Prof. Bermann); Norton 
v. United States, 581 F.2d at 398 (Butzner. J., dissenting). As Deputy Attorney General Civiletti 

[T]he good faith defense doesn't mean that the conduct is not regrettable, terribly 
negligent, that everyone doesn't feel awful remorse about it. and there hasn't been real 


As for the claim that a determination of hability absent bad faith would 
unfairly stigmatize the defendant official, it must be remembered that in such a 
case the official did commit a constitutional injury, even though his conduct 
may have been unintentional. ''^ Surely solicitude for the tortfeasor's reputation 
cannot be allowed to override his victim's right to compensation. And surely 
the fact of the tortfeasor's good faith cannot be allowed to remove the basis for 
disciplining him on grounds other than motive, such as negligence or incom- 
petence. There is no reason why an official should not be held accountable for 
such misfeasance.-^^ 

Congress, of course, may specifically provide by legislation that no finding 
of liability on the government's part in a constitutional tort case shall give rise 
to any adverse inference as to the official's good faith. Indeed, it would appear 
sound for Congress to provide that, once the United States has been substituted 
for an official in a constitutional tort suit, the official shall be free to treat that 
suit — say, in answering questions of potential creditors, employers, or others — 
as if he had never been named as a defendant. Congress should specifically 

harm or substantial harm, if not physically, to money values or at least to the dignity of 
the individual and the privacy of the individual. 

I will give you a quick example. Investigators, three of them, go to a home of suspects 
looking for racketeering — numbers, or whatever. When they go in the home and they are 
queried about it later, each one says he thought the other had a search warrant. None of 
them had a search warrant. 

They find a man and woman 75 years old in there. They take the woman and strip the 
woman down and do cavity searches and whatever — in the wrong home. 

That is not a gift when your are talking about compensation for that kind of wrong under 
the fourth amendment. 

1979 House Hearings, supra note 4, at 19. 

35. See supra note 29. As Deputy Attorney General Civiletti pointed out in his testimony, 
id., the good-faith defense is applicable not only when the relevant law was unclear or not yet 
established, but when other circumstances may have operated to render a violation of constitutional 
rights unintentional or unknowing. Moreover, in the context of common-law torts, to excuse a wrong 
it is ordinarily not enough a show that the relevant law had not previously been declared. 

36. Public Citizen criticized H.R. 7034 for its original limited retention (in H.R. 24) of the 
good-faith defense with respect to the conduct of Members of Congress, judges, prosecutors, or 
others performing analogous functions. 1982 House Hearings, supra note 1 at 357-58 (statement 
of Alan B. Morrison, Director of Litigation, Public Citizen). Presumably, new legislation proceeding 
along those lines would now provide for retention of the defense with respect to the conduct of the 
President as well. See Nixon v. Fitzgerald, 102 S. Ct. 2690, 2705 (1982). Nevertheless, there would 
appear to be some merit to Mr. Morrison's contention that "all the reasons that lead to the elimination 
of the good faith defense in all other situations apply here." 1982 House Hearings, supra note I, 
at 357 (Morrison statement). If the concern is to spare such officials the ordeal of administrative 
proceedings in the event that liability is established, future legislation might be drafted to bar any 
administrative investigation or proceeding against such officials upon a showing that an otherwise 
available immunity could have barred recovery in a given case in which liability was found. The 
power of Congress to abrogate the absolute or qualified immunity of any federal official, even 
without transferring liability to the United States, seems clear. See Nixon v. Fitzgerald, 102 S. Ct. 
at 2719 n.27. 


provide that the defendant official is entitled to assert a good-faith defense in 
any administrative investigation or disciplinary proceeding under the bill.^^ 

Apart from the respects in which waiver by the government of the good- 
faith defense would vindicate the goals of compensation and deterrence, waiver 
of the defense would also operate to limit discovery and facilitate the disposition 
of claims. ^^ As Elliot Richardson, otherwise a proponent of good-faith immunity, 
has noted, "retention of the good faith defense [would] cause federal officials 
to continue to be enmeshed in years of pre-trial depositions and document dis- 
covery aimed at determining the state of mind of an official. ""^^ Reducing con- 
stitutional tort suits to these two issues — did a violation of the plaintiff's 
constitutional rights occur, and was that injury caused by a government official 
acting within the scope of his office or employment — would dramatically sim- 
plify trials, and encourage early settlement. "^^ 

One other argument is sometimes advanced in favor of retaining the good- 
faith defense for the government in constitutional tort suits — namely, that waiver 

37. H.R. 7034, § 202 (proposed 28 U.S.C. § 2700). As Deputy Assistant Attorney General 
Jaffe noted in connection with legislation supported by the Carter Administration, "in any disciplinary 
proceeding, the good faith of the employees will be very much an issue, and if (a defendant employee] 
can establish that he had a good faith belief that what he was doing was lawful and proper, then 
disciplinary proceedings on that basis would not proceed." 1978 House Hearings, supra note 1, at 

38 . As the Justice Department observed in commenting on the cost impact of earlier legislation 
providing for waiver of the good-faith defense: 

Major discovery savings would . . . accrue to the agencies and departments em- 
ploying those individuals who have been sued. Currently, the burden on an agency can 
be immense. The FBI, for example, has had to examine hundreds of thousands of documents 
in connection with suits against individuals. In addition to the savings associated with the 
discovery process, simplification of the issues involved in a constitutional tort action will 
reduce the amount of attorney time necessary for each case. 

7979 House Hearings, supra note 4, at 13 n.l (Department of Justice Memorandum of May 7, 

Although Harlow v. Fitzgerald may have somewhat simplified trial of the good-faith issue with 
respect to the question of unclear applicable law, see supra note 24, other subjective elements would 
certainly remain to be asserted or denied in establishing an official's state of mind, and thus the 
problem of discovery would continue to plague constitutional tort suits where the good-faith defense 
could otherwise be raised. 

39. 1982 Senate Hearing, supra note 1, at 125. Mr. Richardson appeared to suggest that this 
problem should be treated by removing the subjective element from the good faith test. Id., as the 
Supreme Court did in Harlow with respect to unclear applicable law. 

40. The strength of the official's good-faith defense has been an important element in settle- 
ment discussions. 1982 House Hearings, supra note 1, at 251 (testimony of Royce C. Lamberth, 
Chief, Civil Division. Office of the U.S. Attorney for the District of Columbia). Whatever the 
impact of Harlow on the discovery involved whenever the good-faith defense is asserted, see supra 
note 38, the assertion of the good-faith defense may be expected frequently to postpone settlement 
pending appellate review of a district court's ruling on summary judgment for either party with 
respect to the so-called "objective" factors upon which the good-faith defense must now rely. 


of the defense would enable plaintiffs to "prevail more easily, thereby increasing 
economic costs to the [glovernment.'"^' 

As to this argument, it must be observed at the outset that any increased 
cost to the government by virtue of its expanded liability will at least partially 
be offset by the litigation savings that would result from simplification of the 
issues, "^^ and the incentive for early settlement. "^^ And certainly the "economic 
costs to the government" may be moderated by setting liquidated damages in a 
reasonable amount/'* But there is a deeper, more fundamental reason for allowing 
recovery notwithstanding cost. As the current Administration has forthrightly 

[Wjhere someone has a genuine grievance — where someone has a genuine 
claim that he has been deprived of a constitutional right — we ought to afford 
him avenues of relief without regard to cost. We cannot put a price on 
constitutional liberties. "^^ 

If a constitutional injury has occurred, redress should not be denied. 

(b) The Search for an Effective Substitute Deterrent 

As noted above, a key source of dissatisfaction with the existing system of 
civil sanctions is its supposed "overdeterrent" effect — viz., that the threat of 
personal liability (or, at a minimum, the risk of being sued) deters not only 
improper conduct by government officials, but proper conduct as well."^^ On the 
other hand, the Bivens remedy has been made available precisely because, stand- 
ing alone, the remedy of suing the United States "is not a sufficient protector 
of the citizens' constitutional rights. '"^^ 

Indeed, in stressing the salutary deterrent function served by the threat of 
personal liability — including punitive damages — the Supreme Court has specif- 
ically emphasized "the doubt cast on the validity of the assumption that there 
exist adequate mechanisms for disciplining federal employees" who violate 
constitutional rights. "^^ The Court has implied that the threat of disciplinary action 

41. 1982 House Hearings, supra note 1, at 138 (statement of Joseph A. Morris, General 
Counsel, Office of Personnel Management (0PM)). 

42. See supra note 38. 

43. See supra note 40. 

44. As noted above, liquidated damages under H.R. 7034 would have been set at the greater 
of $2,000 or, in the case of a continuing violation, $200 per day for each such violation. See supra 
note 12. Liquidated damages under S. 1775 would have been set at the greater of $1,000 or, in the 
case of a continuing violation, $100 per day for each day up to a maximum of $15,000. S. 1775, 
§ 3 (proposed 28 U.S.C. § 2674(b)(2)). 

45. 1982 House Hearings, supra note 1, at 141 (statementof Joseph A. Morris, 0PM) (arguing 
that possibility of increased damage awards by virtue of substitution of government as defendant in 
constitutional tort suits is worth risking). 

46. See supra note 2. 

47. Carlson v. Green, 100 S. Ct. at 1474. 

48. Id. at 1473 & n.8, citing 1978 House Hearings, supra note 1 , at 6 (statement of Attorney 
General Bell). As the existence of the exclusionary rule demonstrates, existing mechanisms of 
deterrence have not been deemed sufficient to deter Fourth Amendment violations. See Mapp v. 
Ohio, 367 U.S. 634 (1961). If the Supreme Court should eventually adopt a "good faith" exception 


under existing mechanisms is, at most, a supplement to, and not a substitute 
for, the deterrence imposed by the risk of personal liability.'*^ 

One basic question, then, in replacing the existing system of civil sanctions 
with a system of "enterprise liability" is how to assure that the substitute 
deterrent mechanism will equally serve the goal of measured deterrence. 

The approach taken in H.R. 7034 and S. 1775 was one of bare reliance on 
existing mechanisms. The two bills, as noted previously, provided that, when 
there has been a judgment against or money settlement by the United States, the 
Attorney General "shall forward the matter for such further administrative in- 
vestigation or disciplinary action as may be appropriate" to the head of the 
defendant official's department or agency. ^^ 

This provision was criticized by many for failing to offer an adequate 
substitute for the deterrence function of the existing system of civil sanctions — 
in marked contrast to earlier proposals, whose somewhat cumbersome provisions 
for administrative action upon determination of liability, and for judicial review 

to the exclusionary rule, the Court may well signal Congress that additional, substitute mechanisms 
of deterrence are needed. 5f^ Transcript of Oral Argument in Illinois v. Gates, No. 81-430 
(Feb. /Mar.. 1983) (question of Justice Stevens during Reargument). 

49. Carlson v. Green, 100 S. Ct. at 1473. Much of the doubt about the efficacy of existing 
mechanisms of accountability is attributable to the law enforcement and intelligence agency abuses 
revealed during the last decade. See 1979 House Hearings, supra note 4, at 45 & n.32 (Statement 
of Karen Christensen, Legislative Counsel, American Civil Liberties Union). See also "Prosecution 
for Domestic Spying Rejected in 1977." New York Times, Mar. 15, 1982, at AI2, col. 1 (describing 
contents of 1976 report by Justice Department acknowledging apparent violation of "fundamental 
constitutional rights" by federal agencies and officials, but recommending against criminal prose- 
cution); Association of the Bar of the City of New York, Committee on Civil Rights, Intelligence 
Agency Abuses: The Need for a Temporary Special Prosecutor, 31 Rec. A.B. City N.Y. 601 (1976); 
U.S. Department of Justice, Report Concerning Investigation and Prosecutorial Decisions with 
Respect to Central Intelligence Agency Mail Opening Activities in the United States (1977). 

One witness noted in hearings on H.R. 2659 the Federal Bureau of Investigation's administrative 
inquiry into illegal investigative activities directed against the Weather Underground: 

In April. 1979 the Department of Justice Office of Professional Responsibility informed 
FBI Director William Webster that it had evidence of 32 illegal surreptitious entries, 17 
illegal wiretaps, 2 unauthorized microphone installations, and numerous illegal mail open- 
ings conducted against relatives and associates of Weather Underground members. After 
eight months of review, the Director informed the Attorney General that of the 61 special 
agents and 7 supervisors implicated in these illegal activities, disciplinary action was 
appropriate for only 2 agents and 4 supervisors. Ultimately, only mild sanctions were 
applied to the six employees. Letter from William H. Webster, Director, Federal Bureau 
of Investigation to Attorney General Griffin B. Bell at 1 (Dec. 5, 1978). Two street agents 
were 'censured,' meaning that they merely received a letter which identifies a deficient 
act -or omission and calls for proper conduct in the future. Of the [four] supervisors, two 
were 'recommended for dismissal.' one was demoted and the fourth was suspended for 
30 days without pay. Dismissal does not affect vested retirement or pension rights. 

7979 House Hearings, supra note 4, at 45 & n.2 (statement of Karen Christensen, ACLU) (text 
and footnotes conjoined). 

50. H.R. 7034, § 202 (proposed 28 U.S.C. § 2700); S. 1775, § 5(c) (proposed 28 U.S.C. § 


of such action, may well have accounted for that legislation's failure to be 

Plainly some middle ground must be found. An administrative mechanism 
that depends on an agency's inclination to repudiate the conduct of its officials 
can scarcely be expected to foster public confidence that errant officials will 
either be made to account for their actions, or will be effectively dissuaded from 
future misconduct. ^^ On the one hand, an agency may be reluctant to punish 
officials who have committed wrongful acts in the belief that "they were doing 
their duty to their country and to their agency. "^^ At the same time, however, 
internal proceedings may afford agency officials an opportunity to use the de- 
fendant official as a scapegoat for their own wrongdoing, or to punish him for 
bringing their misconduct to light in the course of the proceedings.^"^ Indeed, 
there is good reason to suppose that administrative proceedings arising from 
constitutional torts will frequently involve claims implicating agency policy or 
policymakers in the defendant official's misconduct, ^^ and thereby compromise 
the agency's impartiality.^^ 

Thus, with sound reason, the Federal Executive & Professional Association 
has urged that "the proceeding ... not be conducted by the head of the agency 
in which the employee or employees work," but rather that an independent 
agency be designated for that purpose. -^^ Remitting administrative action to such 
an independent body would, by declining to rely solely for discharge of disci- 
plinary responsibilities on those with the most incentive to misuse it, presumably 
meet at least some of the concerns of the previous Administration and such 
groups as Public Citizen, which had argued for victim participation in indepen- 
dent disciplinary proceedings as "essential" to any "meaningful substitute" for 
the existing mechanism of actual and punitive damages. -^^ 

51. See S. 3314, supra note 5, § 13; H.R. 2659, supra note 5, § 8; S. 695, supra note 5, § 
12(a). H.R. 2659 and S. 695 were Carter Administration proposals. 

52. The concerns expressed go beyond the law enforcement and intelligence context. Thus 
Senator Metzenbaum stated, to Attorney General Bell's expression of assent, that: 

[I]n the case of employees [accused of wrongdoing], you have to have some kind of a 
review board that is totally unresponsive to the agency that is itself involved in the alleged 
wrongdoing. Or else the person who claims to have been wronged will never feel that 
they got justice; and probably the American people, when they learn about it, will not 
think they got justice. It will always be suspect — whether it is the CIA investigating its 
own people, or Cointel, or whether it is HEW or any other agency. 
1978 Senate Hearing, supra note 4, at 13. 

53. 7979 House Hearings, supra note 4, at 127 (remarks of Rep. McClory). 

54. Id. 

55. Id. Indeed, U.S. Attorney Harris specifically cited the tendency of defendant officials to 
plead that they were only "following . . . orders." 1982 House Hearings, supra note I, at 242. 

56. See supra notes 46 & 51. 

57. 1982 House Hearings, supra note 1, at 330-31 (statement of Richard Pelz of FEPA). 

58. 1982 House Hearings, supra note 1, at 361 (statement of Alan B. Morrison); see 1978 
Senate Hearing, supra note 4, at 6 (statement of Attorney General Bell) (advocating "procedures 
in which the injured person can participate in a meaningful way"). In its testimony on H.R. 24, 
Public Citizen called for: 


One approach along these lines would be to establish, for independent 
investigation and, in appropriate cases, prosecution of disciplinary proceedings, 
an Office of Disciplinary Counsel modeled on the existing Office of Special 
Counsel under the Civil Service Reform Act of 1978.^"^ Another approach — one 
advocated by the Federal Executive & Professional Association ("FEPA") — 
would be to utilize the Merit Systems Protection Board ("MSPB") itself as a 
disciplinary body.^ As the PEP A observed, the Special Counsel himself is 
already authorized to investigate abuses of authority in the context of prohibited 
personnel practices.^' 

Thus Congress could specifically require the Special Counsel to investigate 
constitutional tort cases in which there had been judgment of liability or money 
compromise by the government, and direct the prosecution of such cases, when 
appropriate, before the Merit Systems Protection Board. No judicial determi- 
nation in the constitutional tort suit would be binding on the defendant official 
in proceedings before the Merit Systems Protection Board, and the defendant 
official could assert that all defenses that would have been available to him had 
he been a party to the constitutional tort suit under the old system of civil 
sanctions. ^^ 

( 1 ) the right of the victim to initiate an administrative investigation which cannot be terminated 
without adequate reasons; 

(2) the right of the victim to participate in an appropriate manner in the investigation and 
subsequent disciplinary proceedings, if any; and 

(3) the right of the victim to agency, and ultimately judicial, review if no punishment is meted 
out, or if the punishment is wholly inadequate for the violation. 

1982 House Hearings, supra note 1 , at 361 (statement of Alan B. Morrison). These recommendations 
were incorporated in legislation supported by the previous Administration. See Bell, supra note 4, 
at 12-13. 

59. Act of Oct. 10, 1978, Pub. L. No. 95-367 § 202, 92 Stat. 1111 (codified at 5 U.S.C. § 
§ 1201-1209). This option — establishing an Office of Disciplinary Counsel — was proposed by the 
Fund for Constitutional Government. See 1979 House Hearings, supra note 4, at 74. 

60. 1982 House Hearing, supra note 1, at 31 1 (testimony of Richard K. Pelz), citing 5 U.S.C. 
§ 1207 (authorizing MSPB to impose disciplinary sanctions). The MSPB is authorized, inter alia, 
to hear and adjudicate allegations of prohibited personnel practices, 5 U.S.C. § 1205(a), and to 
enforce compliance with its orders by any federal agency or official. Id. § 1205(a)(2). Compliance 
may be enforced by orders that salary payments be withheld pending compliance. Id. § 1205(d)(2). 

61. 1982 House Hearing, supra note 1. at 31 1 (testimony of Richard Pelz, FEPA), citing 5 
U.S.C. § 1207. Chapter 12 of Title 5 of the United States Code, 5 U.S.C. §§ 1 201 -1 209— pertaining 
to the Merit Systems Protection Board and Special Counsel — are reproduced in Appendix A-2. 

The Special Counsel is currently authorized to receive and investigate allegations of ** prohibited 
personnel practices." 5 U.S.C. § 1206(a)(4). If the Special Counsel determines that there is "a 
substantial likelihood . . . [of] a violation of any law, rule, or regulation, or mismanagement, gross 
waste of funds, abuse of authority, or substantial and specific danger to the public health and safety," 
id. § 1206(a)(3)(A), the Special Counsel may under certain circumstances require the agency head 
to conduct an investigation and submit a written report. Id. § 1206(a)(3)(B). The Special Counsel 
may recommend as to what corrective action should be taken, if any, id. § 1206(c)(1)(A), and, if 
the agency has not taken the corrective action recommended, the Special Counsel may request that 
the MSPB order such corrective action. Id. § 1206(c)(1)(B). 

62. Whatever one's view of any particular option for establishing an administrative mechanism 
with "teeth," it cannot be the case that "permitting the alleged [sic] tort victim to have a say in 
the disciplinary proceedings will 'wipe out' the protection sought to be given [by the new system] 


The agency that employs the official whose conduct has been challenged 
would, of course, continue to be responsible in the first instance for investigating 
and, where appropriate, disciplining the official or implementing other corrective 
steps even prior to the outcome of the constitutional tort suit against the gov- 
ernment. Existing disciplinary mechanisms are directed toward conduct related 
to job performance and are designed to promote administrative efficiency. It 
would be desirable for Congress to provide explicit authorization for applying 
existing disciplinary mechanisms to official misconduct. 

Congress could also provide that, before proceeding with disciplinary action 
against any official for a constitutional tort, the Special Counsel shall determine 
whether the agency that employs the official has already acted, and, if so, whether 
the agency's action is satisfactory from the standpoint of the system's overall 
goals of deterrence and accountability. 

The Special Counsel would initiate disciplinary proceedings before the MSPB 
only if he determined that the agency that employed the official had taken no 
action, or that the agency's action was insufficient; any discipline later imposed 
by the Merit Systems Protection Board would be reduced by any discipline 
imposed by the agency in its own proceedings. The Special Counsel would 
annually compile and forward to the House and Senate Judiciary Committees a 
report setting forth the cases he has reviewed, explaining with particularity why 
he chose to proceed or not to proceed with independent action in each case. 

(c) Jury Trial and ' 'Additional' ' Damages 

In rejecting the argument that the Federal Tort Claims Act provides as 
effective a remedy as Bivens-Xy^t actions, the Supreme Court has specifically 
stressed the availability of punitive damages and the option of a jury trial in 
Bivens suits. ^^ As the Court noted: 

Punitive damages are 'a particular remedial mechanism normally available 
in the federal courts,' 5/v^ai^, 403 U.S. at 397, and are especially appropriate 
to redress the violation by a government official of a citizen's constitutional 
rights. Moreover, punitive damages are available in 'a proper' § 1983 action, 
Carey v. Piphus, 435 U.S. 247, 257 n.ll (1978) (punitive damages not 
awarded because district court found defendants 'did not act with a malicious 
intention to deprive respondents of their rights or to do them [ some] other 
injury'), and Butz v. Economou, supra, suggests that the 'constitutional 
design' would be stood on its head if federal officials did not face at least 

and place the employee in a very tenuous position." 1978 House Hearings, supra note 1, at 32 
(statement of John S. McNemey, National President, Federal Criminal Investigators Association). 
It is difficult to see how authorizing the Special Counsel to act on the victim's complaint, for 
example, or to summon the victim to appear as a witness, would place the defendant official at any 
unfair disadvantage. In any event, even under existing administrative mechanisms, the victim of 
alleged misconduct may trigger an investigation. See, e.g., 5 U.S.C. § 1206. 

63. Carlson v. Green, 100 S. Ct. at 1473-74. Jury trial is currently provided under the FTCA 
only in civil actions against the United States for the recovery of internal revenue taxes erroneously 
of illegally assessed or collected, penalties collected without authority, or any other sum alleged to 
have been excessive or in any manner wrongfully collected under the internal revenue laws. 28 
U.S.C. § 2402. 


the same liability as state officials guilty of the same constitutional transgres- 
sion. 438 U.S. at 504. But punitive damages in an FTCA suit are statutorily 
prohibited. 28 U.S.C. § 2674. Thus FTCA is that much less effective than 
a Bivens action as a deterrent to unconstitutional act.^ 

The Court noted that "after Carey punitive damages may be the only significant 
remedy available in some § 1983 actions where constitutional rights are mali- 
ciously violated but the victim cannot prove compensable injury. "^^ (Such a 
problem, of course, would not arise under legislation providing — as did H.R. 
7034 and S. 1775 — for liquidated as well as actual damages.) 

With respect to the argument that juries have been biased against Bivens 
defendants, the Court answered that no suggestion had been made that judges 
would be more receptive, and the Court noted further that no explanation had 
been given for "why the plaintiff should not retain the choice. "^^ In view of 
the minute number of recoveries by plaintiffs in Bivens suits, ^^ the concern about 
jury bias appears to be wholly misplaced. To be sure, however, juries may be 
more ready to award punitive damages against the government than against 
individual Bivens defendants. But any legitimate concern about the risk of ex- 
cessive jury awards should be met by setting a ceiling on the amount of punitive 
damages that may be awarded for a constitutional tort committed with malice, 
and not by denying such damages altogether, or by denying the plaintiff the 
right to a jury trial. 

Needless to say, such damages are more properly cast as "additional" or 
"exemplary" than as "punitive" damages. For the object of assessing such 
damages against the government is not to punish the sovereign, but to express 
sharp social disapproval of constitutional torts committed with malice, and to 
spur the government with special force to minimize the possibility that such 
malicious conduct will recur. ^'^ As FEPA has noted: 

[T]he lawsuit is not simply a device to recompense the injured party for 
the damages which he or she may have suffered at the hands of Federal 
employees who have overstepped the bounds of their authority, in certain 

64. Id. at 1473-74. 

65. Id. at 1473 n.9. 

66. Id. at 1474. 

67. According to Deputy Attorney General Schmults. "several thousand" constitutional tort 
actions had been filed as of November 1981, but only nine had eventuated in money judgments 
against federal employees. 1982 Senate Hearings, supra note 7, at 23. See Bell, supra note 4, at 
2 n.5 (listing 7 cases in which money judgments had been entered for plaintiffs in Bivens actions). 
The number of cases settled by the defendant official is unknown. 

68. In holding on policy grounds that municipalities ought not to be liable for punitive damages 
under 42 U.S.C. § 1983, the Supreme Court emphasized the availability of punitive damages against 
the offending official. City of Newport v. Fact Concerts, Inc., 101 S. Ct. 2748. 2761 (1981). "In 
our view," the Court stated, "this provides sufficient protection against the prospect that a public 
official may commit recurrent constitutional violations by reason of his office." Id. Under a system 
of federal enterprise liability that would make the United States exclusively liable for constitutional 
torts, of course, the availability of punitive damages against an offending official could not be relied 
upon to "directly advance[ ] the public's interest in preventing repeated constitutional violations." 


cases. There is more involved than simple . . . compensation. At issue are 
alleged deprivations of basic and precious individual rights guaranteed by 
our constitution, particularly in the Bill of Rights. The litigation becomes 
an opportunity for the public through the voice of the court to make a 
statement that the abuse of authority was so unjustified as to threaten the 
basic principles of a responsive and responsible government in a democratic 
society. The award of punitive damages would make such a statement. . . . 
[Moreover,] the Attorney General in settling a suit [should be allowed] to 
agree to punitive damages, which would constitute a statement on his part 
that the action of the Federal employee or employees transcended the bound[s] 
of acceptability and excuse. ^^ 

Insofar as the goal of "additional" damages would serve such a function, al- 
lowing such damages to be assessed would complement the Special Counsel 
procedures outlined above. ^^ 

Even more fundamentally, allowing additional damage awards against the 
United States in appropriate cases may be constitutionally required by the prin- 
ciple that Congress may not strip the federal judiciary of the power to fashion 
remedies as deemed necessary to enforce constitutional rights.^' Whether the 
right to jury trial in constitutional tort cases may be eliminated would presumably 
depend, at least in part, on the credibility of the requisite congressional decla- 
ration that the new non-jury trial remedy would be "equally effective" as the 
existing jury trial remedy. ^^ 

Of course, allowing malice to be placed in issue at trial may entail some 
sacrifice of the litigation economies achieved by waiver of the good-faith defense. 
But such economies would not altogether be eliminated, since at least the good- 
faith issue itself would have been placed beyond trial, and, as noted above, there 
are sound reasons for doing that, independent of litigation savings. Moreover, 

69. See 1982 House Hearings, supra note 1, at 328-29 (statement of Richard K. Pelz). 

70. See supra pp. 4-26 to 4-29. 

7L See 1981 Senate Hearings, supra note 7, at 25-6 (Neubome statement). Cf. Jacobs v. 
United States, 290 U.S. 13, 16-17 (1933) (per Hughes, C.J.) (notwithstanding lack of statutory 
provision for award of interest on amount of loss caused by taking, constitutional requirement of 
just compensation held to require inclusion of interest in damage award; "the right to just compen- 
sation could not be taken away by statute or be qualified by the omission of a provision for interest 
where such an allowance was appropriate in order to make the compensation adequate."). See 
generally Tribe, Jurisdictional Gerrymandering: Zoning Disfavored Rights Out of the Federal Courts, 
16 Harv. C.R.-C.L. L. Rev. 129 (1981); Saeger, The Supreme Court, 1980 Term— Foreword: 
Constitutional Limitations on Congress' Authority to Regulate the Jurisdiction of the Federal Courts, 
95 Harv. L. Rev. 17 (1981). 

72. Carlson v. Green, 100 S. Ct. 1468, 1472 (1980). To be sure, H.R. 24, § 3, declares that 
the exclusive remedy against the United States "shall be deemed an equally effective substitute for 
any recovery against any employee of the United States for tort claims arising under the Constitution. ' ' 
Although the Supreme Court in Carlson stated that a Bivens-type action may be defeated by such 
a declaration, 100 S. Ct. at 1472, the validity of the declaration itself would surely be subject to 
judicial review. It is far from certain that a mere ipse dixit by Congress would survive review. 


the expense of litigating allegations of malice can be expected, at least until 
experience proves otherwise, to discourage their frivolous assertion. ^^ 

(d) Attorney Fees 

As noted above, H.R. 7034, but not S. 1775, provided that a prevailing 
plaintiff (i.e., one who has won a judgment, or received a cash settlement from 
the government) would be entitled to "a reasonable attorney's fee and other 
litigation costs reasonably incurred, including attorney fees or costs attributable 
to processing [the initial] administrative claim. . . ."^"^ This provision, too, has 
been a source of controversy. 

Thus, on the one hand, the Association of Federal Investigators and the 
SEC favored eliminating attorney fees altogether, arguing that the possibility of 
such fees would only encourage litigation and, even worse, induce "artful plead- 
ing" by lawyers to set forth constitutional rather than non-constitutional tort 
claims whenever possible. ^^ On the other hand, the same considerations that 
support awarding attorney fees under 42 U.S.C. § 1988 in cases of proven 
violations of constitutional rights would seem to apply here as well. Inasmuch 
as the attorney fee issue raises questions that go far beyond the constitutional 
tort context, however, any recommendation on the issue would be beyond the 
scope of this report. 

(e) Torts of Former Officials 

Although H.R. 7034 and S. 1775 appeared to cover claims for acts or 
omissions of former government officials committed while such officials were 
in government service, the bills provided no administrative mechanism for bring- 
ing former officials themselves to account. Under H.R. 2659, a Carter Admin- 
istration bill, former officials could elect either to be sued individually after 
leaving government service, or to have the government substituted in their place. 
In choosing the latter, the former official would have agreed to submit to a 
disciplinary proceeding that could have resulted in a fine equal to as much as 
one-twelfth of the annual salary earned at the time the act or omission occurred. ^^ 
Whatever one may think of the specific remedy authorized by H.R. 2659, it 
would seem wholly anomalous to place beyond accountability officials who had 
left government service before a determination of liability. 

(f) ' 'Scope of Office' ' versus ' 'Color of Law' ' 

A serious issue is presented by the Attorney General certification require- 
ment under H.R. 7034 and S. 1775. First, as a matter of public policy it would 

73. To minimize the discovery burden if punitive damages are made available, the trial might 
proceed in two stages, in the first determining liability, and in the second — involving discovery on 
the issue of malice — following the first only if liability is established. 

74. H.R. 7034, § 202 (proposed 28 U.S.C.§ 2697(a)). 

75. 1982 House Hearings, supra note 1, at 341 (statement of J. Brian Hyland); id. at 284 
(statement of SEC); see id at 258 (statement of William H. Taft IV, General Counsel, Department 
of Defense). 

76. See 1979 House Hearings, supra note 4, at 81 (statement of FBI Director Webster) 
(praising former official provision as "fair and worthwhile"). 


appear advisable to replace the ''scope of office or employment" criterion with 
the broader requirement of ''color of law," tracking the criterion of liability 
applicable to state and local officials under 42 U.S.C. § 1983.^^ 

By expanding the circumstances under which constitutional violations would 
be judicially cognizable, such a substitution would obviously enable an enterprise 
liability system to promote even more effectively its twin goals of assuring victim 
compensation and deterring official misconduct. As Professor Hermann has tes- 
tified, this approach: 

[I]s more consistent with the values of full compensation and loss-spreading 
for injuries somehow connected with governmental action. It would more 
effectively relieve the official of the prospect of personal litigation and 
liability, which is one of the chief purposes behind H.R. 2659. Finally, it 
would often avoid the necessity of premature threshold determinations on 
the question whether the employee was acting within the scope of his 
authority or solely under color of office. Both in terms of its relevance and 
its feasibility, this determination is best left to the disciplinary phase of the 
affair. ^^ 

Nevertheless, before exposing the United States to such liability — vastly greater 
than under the Bivens doctrine — prudence counsels a more limited experiment, 
establishing liability of the United States only where an official has committed 
a constitutional tort within the scope of his office or employment. 

(g) The Problematic Distinction Between Constitutional and Non- 
constitutional Torts 

The distinction between constitutional and non-constitutional torts — a dis- 
tinction that would have been enshrined in both H.R. 7034 and S. 1775^^ — is 
significant insofar as the Bivens remedy against individual government officials 
provides plaintiffs with rights not afforded plaintiffs in FTCA suits against the 
government — to wit, the right to jury trial and punitive damages. ^° In addition, 
the right of a successful plaintiff to attorney fees in constitutional tort cases — 
by analogy to 42 U . S . C . § 1 988 — is also recognized in some legislative proposals 
to replace the Bivens remedy but is not currently afforded a successful plaintiff 
under the FTCA.^' 

77. This was the approach embodied by H.R. 2659, supra note 4, the Carter Administration 
bill, see 1979 House Hearings, supra note 4. at 8 (statement of Deputy Attorney General Civiletti); 
1978 Senate Hearings, supra note 4, at 39-40 (S. 2117), as well as by eariler legislative proposals. 
See S. Rep. No. 588, supra note 22, at 34; S. Rep. No. 755, supra note 4, at 337-38. 

78. 7979 House Hearings, supra note 4, at 124. Under H.R. 7034 and S. 1775, an individual 
government official would remain personally liable for constitutional misconduct committed outside 
of the scope of his office or employment, but under color of law. These bills do not, and could not, 
purport to provide an "equally effective" substitute for the existing system of treating such mis- 
conduct. Bivens v. Six Unknown Named Agents, 403 U.S. at 397. 

79. See supra note 1 1. 

80. See Carlson v. Green, 101 S. Ct. at 1413-14. 

81. See, e.g., H.R. 7034, § 202 (proposed 28 U.S.C. § 2698(a)). Cf. Lauritzen v. Secretary 
of the Navy, No. CV 81-879 (AWT) (CD. Cal., Sept. 9, 1982) (available Oct. 15, 1982, on 
LEXIS, Genfed library, Dist. file) (holding that successful Bivens plaintiff may recover attorney 
fees from the United States under § 204(a) of the Equal Access to Justice Act, 28 U.S.C. § 2412). 


The availability of such "extra" rights has been understood to render the 
Bivens remedy "more effective" than the FTC A remedy. ^^ Undoubtedly for this 
reason, the distinction is maintained in proposed legislation to replace the Bivens 
remedy. The alternative to maintaining the distinction would be either to extend 
such "extra" rights to non-constitutional suits, or to eliminate them from "con- 
stitutional" tort suits. ^^ 

The practical problem entailed by the distinction, of course, is its natural 
tendency to encourage "skillful counsel [to] plead the existence of a constitutional 
tort when their case in reality sounds in a traditional, common law cause of 
action."^"* If the distinction is to be maintained in any legislation replacing the 
Bivens remedy — as this report recommends it ought to be — Congress should 
make clear in the legislative history its intention that the courts be alert to the 
possibility of such skillful pleading, and decide a plaintiff's entitlement to the 
special advantages of pleading constitutional torts "based on the true gravamen 
of the tort as alleged and proven. "^^ 

B . Recommendation 

Under current law, individual federal officials may be held personally liable 
for constitutional violations they are found to have committed while acting within 
the scope of their office or employment. Damages may not be recovered against 
the United States for violations of constitutional rights as such, although claims 
arising out of the same conduct may sometimes be stated against the United 
States under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-80. 

There is nearly universal agreement that the existing system of civil sanctions 
for constitutional violations by federal officials neither provides adequate as- 
surance of compensation for victims of such violations, nor affords the degree 
of measured deterrence required to discourage improper conduct by government 
officials without discouraging proper conduct as well. In addition, the federal 
government often has interests at stake in constitutional tort litigation involving 
its officials which cannot adequately be represented by the individual official as 


To serve the primary goals of compensation, deterrence, and fairness in 
dealing with constitutional torts committed by federal officials, and to afford a 
solution to the problems perceived to flow from the current system of individual 
liability. Congress should enact legislation providing that the United States shall 
be exclusively liable for damages for torts arising under the Constitution of the 

82. Carlson v. Green, lOI S. Ct. at 1473. 

83. Compare 1979 House Hearings, supra note 4. at 41 (statement of Karen Christensen, 
ACLU) (urging extension of such rights to non-constitutional tort suits) with 1982 House Hearings, 
supra note 1, at 284 (statement of SEC) (urging elimination of such rights in constitutional tort 

84. U.S. Department of Justice, Section-by-Section Analysis of S. 695, at 8 (1979). 

85. Id. 


United States and committed by federal officials while acting within the scope 
of their office or employment. 

Such legislation should provide: 

(1) That, in constitutional tort actions, the United States may not assert as 
a defense the absolute or qualified immunity of the official whose conduct gave 
rise to the claim, or his reasonable good-faith belief in the lawfulness of his 
conduct. Such immunities, and the good-faith defense, have been judicially 
created for policy reasons to assure that exposure to personal liability shall not 
deter individual officials from the vigorous discharge of their responsibilities, 
and serve no purpose when the government is substituted as defendant other 
than to limit the extent to which genuine victims of constitutional injuries may 
secure redress. In providing that the United States shall not assert such immunities 
or the good-faith defense. Congress may wish to provide that the United States 
shall be permitted to assert any applicable immunities of the President, and, 
perhaps, of federal judges and members of Congress. 

(2) That there shall be vested in an office of the Executive Branch inde- 
pendent authority to investigate constitutional tort cases in which there has been 
a judgment of liability against or money compromise by the United States, and 
that such office shall be vested with independent authority to conduct disciplinary 
proceedings in such cases as may be appropriate. Such a disciplinary mechanism 
is essential to perform the deterrent and corrective functions now served by the 
damage action remedy against individual government officials; existing admin- 
istrative disciplinary mechanisms, by themselves, are neither directed toward 
nor capable of performing these functions under current law. 

(a) One approach for implementing this recommendation would be to es- 
tablish an Office of Disciplinary Counsel to investigate and, where appropriate, 
to prosecute official misconduct before an administrative tribunal independent 
of the agency that employed the offending official. Another approach would be 
to augment the existing authority and responsibilities of the Office of Special 
Counsel and the Merit Systems Protection Board under 5 U.S.C. §§ 1201-09 
to perform the functions described herein. 

(b) If Congress should choose to rely on the Office of Special Counsel and 
the Merit Systems Protection Board in implementing this recommendation, sep- 
arate mechanisms would nevertheless be required to provide for independent 
investigation and disciplinary proceedings with respect to federal officials not 
subject to the jurisdiction of the Office of Special Counsel and the Merit Systems 
Protection Board. 

(c) Notwithstanding the authority of the independent disciplinary mecha- 
nism utilized to implement this recommendation, the agency that employed the 
offending official would be responsible in the first instance for investigation and, 
where appropriate, for disciplining the official or implementing other corrective 
steps, and such action should not be postponed pending the outcome of any 
constitutional tort suit that may be filed against the United States. Congress 
should provide every federal agency with explicit statutory authority to employ 
existing administrative mechanisms for disciplining officials found to have vi- 
olated constitutional rights. 


(d) Before proceeding with disciplinary action against any official for a 
violation of constitutional rights, the Disciplinary Counsel or Special Counsel 
should determine whether the agency that employed the official has already taken 
disciplinary or other corrective action, and, if so, whether the agency's action 
is satisfactory from the standpoint of the system's overall goals of deterrence 
and accountability. The Disciplinary Counsel or Special Counsel would be au- 
thorized to initiate disciplinary proceedings before the Merit Systems Protection 
Board or other independent administrative tribunal only if he determined that 
the agency that employed the official had taken no disciplinary or other corrective 
action, or that the agency's action was insufficient. Any discipline later imposed 
by the Merit Systems Protection Board or other independent administrative tri- 
bunal would be reduced by any discipline imposed by the agency in its own 

(e) The Disciplinary Counsel or Special Counsel should be directed to 
compile and forward to the House and Senate Judiciary Committees periodic 
reports setting forth the cases he has reviewed, and explaining with particularity 
why he chose to proceed or not to proceed with independent disciplinary action 
in each case. 

(3) That Congress should provide not only for actual damages but for 
reasonable liquidated damages in the event that the actual damages are nominal 
or nonexistent because the injury caused by the violation of a constitutional right 
is of an intangible nature. Congress should also consider allowing "additional" 
damages against the United States in cases where the conduct giving rise to the 
tort was undertaken with the malicious intention to cause a deprivation of con- 
stitutional rights or with reckless disregard for the plaintiff's constitutional rights. 

(4) That the right to jury trial should be retained for plaintiffs whose claims 
had arisen as of the effective date of the legislation implementing this recom- 
mendation, and that Congress should consider extending the jury trial right to 
plaintiffs whose claims arose subsequent to the effective date of the legislation. 

(5) That Congress consider the appropriateness of allowing attorney fees 
under such legislation in the context of a more comprehensive review of the 
attorney fee issue in federal legislation across-the-board. 

(6) That legislation implementing this recommendation provide a mecha- 
nism for holding former government officials accountable for constitutional torts 
committed while they were in office, where a plaintiff has secured a judgment 
against or money compromise by the United States. Congress should consider 
allowing federal officials leaving government service to elect either to be sued 
individually for such torts, or to have the United States substituted in their place; 
and, in electing the latter, such officials would agree to submit to appropriate 
sanctions imposed by the agency for which they were employed when their 
tortious conduct occurred. 

(7) That such legislation should provide for periodic review by the House 
and Senate Judiciary Committees of the operation of the legislation implementing 
the new system of federal enterprise liability and independent administrative 


V. Analysis: Changes at the State and Local Level 

In recent years, the Supreme Court has broadened significantly the potential 
liability of state and local officials in cases brought under the Civil Rights Act 
of 1871, 42 U.S. C. § 1983. Similarly, the Court's recent decisions have expanded 
the liability of municipalities and other local governmental units for money 
damages under section 1983 and for the costs of attorney fees pursuant to the 
Civil Rights Attorneys' Fees Act of 1976, 42 U.S.C. § 1988. It is widely believed 
that since section 1988 went into effect the availability of attorney fees in itself 
has contributed greatly to the marked increase of section 1983 cases. These 
developments have been the subject of congressional scrutiny and intense crit- 
icism by state and local officials who claim that the floodgates are open for 
frivolous and expensive lawsuits against state and local officials and munici- 

A substantial number of the section 1983 cases arise from actions taken by 
state and local officials in administering federal assistance programs.^ Federal 
assistance will continue to be a major source of section 1983 litigation because, 
notwithstanding the Reagan Administration's efforts to reduce grant funds, fed- 
eral grant oudays are estimated to exceed $81.4 billion in FY 1983.^ 

And perhaps the greatest irony of the Reagan Administration's "New Fed- 
eralism" initiatives, which are intended to transfer to state and local governments 
greater responsibility for managing federal assistance programs, is that this policy 
could dramatically increase the states' exposure to litigation under section 1983.^ 
For this reason alone, it is appropriate to consider whether improvements can 
be made in the existing civil sanction system to better meet the twin objectives 
of effective management of federal assistance programs by state and local officials 
and the vigorous enforcement of individual rights and federal laws relating to 
such programs. 

This chapter first reviews the major Supreme Court decisions involving 
section 1983 and examines the special problems these decisions present for state 
and local officials and municipalities involved in the administration and man- 
agement of federal assistance. The chapter then presents the results of a nation- 
wide survey of state civil sanction systems and legislative developments. Recent 
congressional proposals to amend sections 1983 and 1988 are also reviewed. 
Conclusions and recommendations are limited to the problems of section 1983 

1 . The term federal assistance as used in this chapter encompasses any disbursement or transfer 
of property — including money — by the federal government that supports programs and projects that 
benefit the public and that is accompanied by an agreement by the recipient to comply with any 
terms or conditions relating to the use of the disbursement or property. 

2. See 0MB , The Budget of the United States Government, 1983, Special Analysis H: Federal 
Aid to State and Local Governments. 

3. Madden & Harkins, New Block Grant Programs Face Period of Adjustment in the Courts, 
V. 4, no. 26, The National Law Journal 29 (March 8, 1982). 


liability insofar as liability arises out of the acts of state and local officials and 
municipalities in administering federal assistance."^ 

A. Liability Under 42 U.S.C. § 1983 

1 . Reprise 

The Supreme Court's 1961 decision in Monroe v. Pape,^ established section 
1983 as the primary basis for the recovery of money damages against state and 
local officials. According to Monroe, section 1983 actions are available to an 
injured party even though a remedy exists under state law, since "the federal 
remedy is supplementary to the state remedy, and the latter need not be first 
sought and refused before that federal one is invoked.''^ However, the Court in 
Monroe retained the concept of absolute immunity from liability for municipal- 
ities under section 1983.^ 

Since Monroe, section 1983 has been used to establish the personal liability 
of several types of public officials and employees at all levels of state and local 
government.^ For example, officials declared amenable to suit under section 
1983 include a state governor,^ state prison officials,'^ township supervisors," 
police officers,'^ a commissioner of police,'^ welfare officials,"^ a school su- 
perintendant,'^ local school officials,'^ and a state commissioner of education.'^ 

4. No attempt is made to discuss and resolve the problem of section 1983 liability in all of 
its complexities. Rather, the discussion of section 1983 liability is limited solely to the somewhat 
relatively narrow area of federal assistance and the special problems for state and local officials and 
municipalities involved with federal assistance programs under that provision. 

5. 365 U.S. 167 (1961). The Court in Monroe considered the * "under color of" state law 
language of section 1983 and held that that section provided "a remedy to parties deprived of 
constitutional rights, privileges and immunities by an official's abuse of his position." Id. at 172. 

6. Id. at 183. In addition, plaintiffs are not required to exhaust state administrative remedies 
prior to bringing a section 1983 action. See Patsy v. Board of Regents, 102 S. Ct. 2557 (1982). 

7. 365 U.S. at 187-92. 

8. States continue to enjoy eleventh amendment immunity to damages suits under section 
1983, since they are not considered to be persons within the meaning of that section. Quern v. 
Jordan, 440 U.S. 332 (1979); Edelman v. Jordan, 415 U.S. 651 (1974). However, as a practical 
matter, a state is subject to liability under section 1983 in cases where injunctive relief against state 
officials in their official capacity is awarded and consequently the state must implement the terms 
of the injunction. Ex parte Young, 209 U.S. 123 (1908). See Edelman, supra, at 667-68 (federal 
courts may enjoin state officials to conform their conduct even though such an injunction may have 
an ancillary effect on the state treasury). 

9. Scheuer v. Rhodes, 416 U.S. 232 (1974). 

10. Procunier v. Navarette, 434 U.S. 555 (1978); cf. Jaworski v. Schmidt, 684 F.2d 498 (7th 
Cir. 1982). 

11. Detz V. Hoover, 539 F. Supp. 532 (E.D. Pa. 1982). 

12. Pierson v. Ray, 386 U.S. 547 (1967); DiGiovanni v. City of Philadelphia, 531 F. Supp. 
141 (E.D. Pa. 1982). 

13. DiGiovanni, supra note 12. 

14. Goldberg v. Kelly, 397 U.S. 254 (1970). 

15. Ingrahm v. Wright, 525 F.2d 909 (5th Cir. 1976). 

16. Wood V. Strickland, 420 U.S. 308 (1975). 

17. Barrea v. Wheller, 475 F.2d 1338 (8th Cir. 1973), af'd, 417 U.S. 402 (1974). 


State and local officials may defend themselves from liability under section 1983 
by asserting a qualified-immunity defense.'^ 

The impact of section 1983 on the activities of local governments and their 
officials increased dramatically with two landmark rulings by the Supreme Court: 
M one 1 1 V. Department of Social Service s^"^ and Owen v. City of Independence.-^^ 
These two decisions broadened the liability of municipalities for the actions of 
their officials in suits under section 1983.^' In Monell, the Court held for the 
first time that municipalities are subject to suit under section 1983, overruling 
a contrary holding made only eighteen years earlier in Monroe.^' The crowning 
blow to local governmental immunity came two years later in Owen, where the 
Court held that local governments could not claim the qualified-immunity de- 
fense. In Owen, the Court ruled that a municipality sued under section 1983 is 
liable under ordinary standards of negligence and may not claim the good faith 
of its officers or agents as a defense to such suits. --^ In reaching this result the 
Court strongly credited public policy considerations of compensation of victims 
and deterrence of future deprivations.'^'^ 

In the aftermath of Monell and Owen, municipalities and other local gov- 
ernmental units^^ face a form of strict liability which closely approaches enter- 
prise liability, since in addition to suing a local official under section 1983, a 
plaintiff may simultaneously sue that official's governmental employer. ^^ Local 
liability, however, is limited to actual damages. According to City of Newport 

18. See Harlow v. Fitzgerald, 102 S. Ct. 2727 (1982), modifying Scheuer v. Rhodes, supra 
note 9. 

19. 436 U.S. 658 (1978). 

20. 445 U.S. 622 (1980). 

21. However, a municipality is not vicariously liable for the acts of its agents or employees 
under the doctrine of respondeat superior. 436 U.S. at 691. See also Leonhard v. U.S., 633 F.2d 
599, 622 (2d Cir. 1980). 

22. In Monell, the court determined that a city policy which required pregnant employees to 
take maternity leave after the fifth month of pregnancy violated rights secured by section 1983. The 
Court consciously reversed its holding in Monroe and declared that local governments could be 
liable for violations of section 1983. The Court stated "[if government, in the execution of its] 
policy or custom, whether made by its lawmakers or those whose edicts or acts may fairly be said 
to represent official policy, inflicts the injury [then] the government [can be held] responsible." 436 
U.S. at 684. 

23. The Court considered the scope of municipal liability under section 1983 in the context 
of a suit by a city police chief challenging his dismissal by the city manager without notice of the 
reasons for dismissal or an opportunity to be heard. 445 U.S. at 625-30. 

24. Id. at 651. 

25. County governments are also subject to suit under section 1983 if under the state con- 
stitution the county is not considered to be part of the state for eleventh amendment purposes. See, 
e.g., Edelman v. Jordan, 415 U.S. 651, 668 n.l2 (1974); Holley v. Lavine, 605 F.2d 638 (2d Cir. 
1979), cert, den., 446 U.S. 913 (1980); Knight v. Carlson, 478 F. Supp. 55 (E.D. Cal. 1979). 

26. See, e.g., Garris v. Rowland, 51 U.S.L.W. 2050 (5th Cir. June 24, 1982). 


V. Fact Concerts, Inc.,^^ section 1983 liability does not extend to punitive 
damages. ^^ 

The potential for a dramatic increase in suits against state and local officials 
and municipalities involved in managing federal grant programs is perhaps seen 
most clearly in Maine v. Thiboutot}"^ In Thiboutot, the Court stated explicitly 
what had been the implicit premise of a generation of section 1983 litigation; 
namely, that section 1983 reaches the "deprivation of any rights secured by the 
constitution and laws"^^ of the United States. The Court held that section 1983 
should be read in accordance with its literal language, and thus should be con- 
strued to "broadly encompass violations of federal statutory law as well as 
constitutional law."-^' 

The Court in Thiboutot also had occasion to consider whether attorney fees 
could be awarded in section 1983 claims based on purely statutory violations. ^^ 
The Court, relying once again on the "plain language" reasoning which it had 
applied to section 1983, held that section 1988 authorized the award of attorney's 
fees in "any § 1983 action. "^^ 

A trend toward increased litigation arising from federal grants and assistance 
may be marked by the decision in Middlesex County Sewerage Authority v. 
National Sea Clammers Ass'n?^ Justice Powell, writing for the majority, held 
that private suits under section 1983 may be used to enforce federal statutory 
rights unless Congress has explicitly or implicitly precluded such suits. Thus, 
section 1983 is not available "when the remedial devices provided in a particular 
act are sufficiently comprehensive, [so as] to demonstrate congressional intent 

27. 453 U.S. 247 (1981). The Supreme Court ruled that municipalities are immune from 
liability for punitive damages under section 1983. According to Justice Blackmun, damages for 
punitive purposes are not "sensibly assessed" against a governmental entity, since the traditional 
objectives of punitive damages — retribution, deterrence and punishment — would not be served by 
imposing such liability on a municipality vis-a-vis its blameless or unknowing taxpayers. Id. at 261 . 

28. The decision in Fact Concerts has been welcomed by local officials. See, e.g.. Municipal 
Liability Under 42 U.S.C. § 1983: Hearings Before the Subcommittee on the Constitution of the 
Senate Committee Judiciary. 97th Cong.. 1st Sess. 322 (1981) (statement of Ken Eikenberry) 
(hereinafter cited as Section 1983 Hearings). 

29. 448 U.S. 1 (1980). 

30. 42 U.S.C. § 1983. 

31. 448 U.S. at 4. 

32. Id. at 9. 

33. Id. While the ruling in Thiboutot underscored the scope of sections 1983 and 1988, the 
Court's decision in the companion case of Maher v. Gagne, 448 U.S. 122 (1980), imposed on 
section 1983 defendants the added burden of attorney's fees in cases where a plaintiff prevails through 
a settlement rather than through litigation leading to favorable judgment. Successful defendants may 
also be reimbursed, but the standards are much more stringent than for plaintiffs. See Cass, Damage 
Suits Against Public Officers. 129U. Pa. L. Rev. 1110, 1155 n. 170(1981). For a general discussion 
of the Act, see Witt, The Civil Rights Attorneys' Fees Awards Act of 1976, 13 Urb. Law. 589 
(1981). One lower court has decided, however, that where a statute provides a private cause of 
action, a plaintiff cannot add a claim under section 1983 for the sole reason of obtaining an attorney 
fee award. See Tatro v. Texas, 516 F. Supp. 968 (N.D. Tex. 1981). 

34. 453 U.S. 1 (1981). 


to preclude such suits. "^^ Importantly, Justice Powell also declared that private 
suits may be precluded if the statute at issue fails to establish the kind of "rights, 
privileges and immunities" enforceable under section 1983;-^^ an exception first 
recognized by Justice Rehnquist in Pennhurst State School and Hospital v. 

Although National Sea Clammers may well have been intended to restrict 
the availability of section 1983, its consequence may be to open the door to 
litigation still further. Justice Powell's attempt to articulate two narrow exceptions 
to section 1983 availability — the "exclusive remedy" and "rights" excep- 
tions — could have the anomalous effect of making section 1983 the remedy of 
choice in disputes involving claims under many federal-state cooperative pro- 
grams. ^^ 

In addition to Thiboutot and National Sea Clammers, the greatest impact 
on the subject of official liability as it relates to federal assistance may evolve 
from a long-running dispute over fund termination involving federal defendants. 
The issue in Velde v. National Black Police Ass' n.^"^ was whether federal officials 
may claim absolute immunity from personal liability for allegedly failing to 
enforce statutory conditions of grant funding against state and local government 
agencies charged with discrimination. 

At the end of its 1981 term the Court remanded Velde in light of its decision 
in the controversial case oi Harlow v. Fitzgerald.^^ As discussed in detail else- 
where, the Court held in that case that the President's chief aides are not entitled 
to absolute immunity. 

The ruling in Velde is significant to all types of assistance mechanisms with 
respect to state and local as well as federal officials. In view of the Court's 
apparent refusal to grant absolute immunity, it is now more certain than ever 
that the decisions of government officials under an assistance program, which 
affect the substantive obligations and rights created by such program, will be 
challenged by litigation notwithstanding the ability of officials to raise and ul- 
timately establish a qualified immunity. 

2. Problems Related To Federal Assistance 

This section analyzes the implications of the recent case law for federal 
assistance by focusing on the 1981 block grant programs, "^^ because these mea- 

35. Id. at 20. 

36. Id. 

37. 451 U.S. 1 ( 1981 ). In Pennhurst, the Supreme Court was concerned about the unintended 
consequences that might result if open-ended and "indeterminate" statutory phrases of grant statutes 
are construed as conferring enforceable rights on individuals. Id. at 4. The significance oi Pennhurst 
is considered below in relation to the obligation of state and local grantees to comply with the 
conditions of grant assistance. See infra text accompanying note 52. 

38. This point is analyzed below more fully in connection with the discussion of block grant 
suits under section 1983. See infra text accompanying note 63. 

39. 102 S. Ct. 3503 (1982). 

40. 102 S. Ct. 2727(1982). 

41 . A block grant is a federal program designed to achieve some broad national purpose and 
in which funds are allocated under a formula to a state government for use in a broad functional 


sures are representative of the current movement giving state and local officials 
more responsibility for managing federal grant programs. It should be noted that 
as a matter of law, the courts have not recognized any difference between block 
grants and other federal grant programs/^ 

a. Rights and Obligations Under Section 1983 

Federal grants are presently the central topic of an impassioned debate over 
the proper role of the federal government in our federal system. To one extent 
or another, the nine block grants enacted in 1981 are each aimed at alleviating 
the perceived frustration over the federal government's intrusion into the tradi- 
tional prerogatives of state and local governments."^-^ 

However, it is apparent that several factors could give rise to potential legal 
problems for state and local officials who manage federal grant programs in- 
cluding block grants. "^"^ For example, at a time when many local governments 
and community groups are experiencing financial difficulties,"^-^ fierce compe- 

area at the state's discretion. See Advisory Commission on Intergovernmental Relations, Back in 
Vogue: The Politics of Block Grant Legislation, n. 16. Intergovernmental Perspective, Washington, 
D.C., U.S. Government Printing Office. Spring 1981. The statutory authorizations for the 1981 
block grant programs are contained in the Omnibus Budget Reconciliation Act of 1981, Pub. L. 
No. 97-35, 97th Cong., 1st Sess., 95 Stat. 357 (Aug. 13, 1981) (hereinafter OBRA) which establishes 
block grants in the areas of preventive health, maternal and child health services, alcohol, drug 
abuse and mental health, primary care, social services, community services, low-income energy 
assistance, community development and education. See also Madden, Lessons From Block Revenue 
Sharing, 36 Fed. B.J. 107 (1977). 

42. Ely V. Welde, 451 F.2d 1 130 (5th Cir. 1971). 

43. In an effort to further shift responsibility to the state and local levels, the President's 1983 
budget package proposes to combine approximately 35 social programs into seven new block grants 
and expand three of the block grants enacted in 1981 to include five existing categorical programs. 
See The Budget of the United States Government, 1983, Special Analysis H, supra note 2. 

44. As a separate matter, it should be noted that a recent General Accounting Office report 
entitled Early Observations On Block Grant Implementation, (GAO/GGD-82-79, Aug. 24, 1982), 
suggests that based on 13 states examined by GAO, state assumption of responsibility for the 1981 
block grants has proceeded rather smoothly. 

45. The declining fiscal condition of American cities of all sizes and the adverse impact on 
cities of federal budget reductions under grants has been well-documented. For example, a November, 
1981 report by the United States Conference of Mayors (USCM) entitled. The FY82 Budget and 
the Cities: A Hundred City Survey, indicates that cuts in federal spending have fallen dispropor- 
tionately on grants to state and local government, with cities being forced to absorb major reductions 
in housing, community development, education, transit, employment and other vital programs. As 
a result, cities have had to reduce substantially both the level and types of services they provide to 
urban residents. 

In addition, a staff study prepared by the Joint Economic Committee of Congress entitled. 
Emergency Interim Survey: Fiscal Condition of 48 Large Cities (January 1982), supports the findings 
of the USCM. The study concludes that a majority of the 48 cities surveyed have had to reduce real 
service expenditure levels for virtually every service offered in order to accommodate reductions in 
federal assistance. The study emphasizes that cities, unable to maintain current service levels, are 
in no position to assume additional administrative or fiscal responsibilities since city governments 
simply do not have the resources to undertake such responsibilities. The survey thus makes the 
gloomy prediction that "there will be less success in making state and local government responsible 
for managing and financing many programs now funded by the Federal Government.' 




tition can be expected between local entities for available funds/^ In addition, 
the unprecedented shift in power to the states to distribute funds received under 
block grants is likely to generate legal challenges to the use and distribution of 
limited funds/^ 

The procedural requirements and substantive conditions imposed on state 
and local grantees under block grant programs are also likely to provide a source 
for legal challenges by third parties and intended beneficiaries'^^ of block grant 
funds. And importantly, state and local grantees will be subject to the so-called 
crosscutting requirements of the more than 20 statutes that govern the expenditure 
of all types of grant funds. "^^ In this regard, state officials are required to make 
contractual assurances that they will perform the various statutory conditions 
which accompany block grant assistance and that they will distribute funds 
equitably to local entities. ^^ 

The assurances made by a state in its annual application will establish 
important rights for the millions of third parties identified by Congress as the 

46. In its April, 1982 report entitled Briefing Book on Block Grants and New Federalism, 
the Coalition on Block Grants and Human Needs emphasizes that one of the impacts of block grants 
on states will be "intense political pressure about allocation choices." The Coalition reports that 
competing groups that once vied for federal appropriations will now devote their attention to allocation 
decisions by state and local officials, and that this competition will become especially "intense" as 
additional funding cuts in grant programs are enacted by Congress. 

47. Although to data there has been relatively little litigation involving the 1981 block grants, 
it is widely acknowledged that the opportunity for extensive litigation is likely to increase as states 
make greater use of their discretion to make funding decisions and as constituencies become frustrated 
by service cuts or other decisions associated with state and local implementation of block grant 
programs. See, e.g.. Coalition on Block Grants and Human Needs, Briefing Book on Block Grants 
and New Federalism, at 71 (April 1982); Remarks of Milton J. Socolar, General Counsel of the 
General Accounting Office, before The Brookings Institute Seminar on New Federalism, Washington, 
D.C., June 21-22, 1982. 

48. Grant programs typically create benefits not only for the direct recipient of the grant, 
usually a state or local government or agency, but also for persons or groups — generally referred 
to as third parties or intended beneficiaries — who often are given certain enumerated rights under 
the statute authorizing the grant program. 

49. There are over sixty government- wide national policies and administrative requirements. 
These policies include, among others, non-discrimination, environmental protection and energy 
conservation, and are reflected in these statutes, including: Title VI of the Civil Rights Act of 1964, 
42 U.S.C. § 2000d et seq.; the Clean Air Act of 1972, 42 U.S.C. § 1857; Title IX of the Education 
Act Amendments of 1972, 20 U.S.C. § I68I et seq.; and the Federal Water Pollution Control Act 
of 1948, 33 U.S.C. § 1251 etseq. GAO's report. Early Obsenations On Block Grant Implementation, 
emphasizes that states have been given little guidance on and have limited experience with federal 
crosscutting requirements. GAO thus suggests that the Office of Management and Budget, in co- 
ordination with federal agencies, develop a program for assessing state compliance with federal 
crosscutting mandates. GAO/GGD-82-79, p.p. 42-44. 

50. The 1981 block grants require each state to submit an application for assistance in which 
the state makes assurances that it will comply with the conditions of federal block grant assistance. 
For example, the statute authorizing the Community Services Block Grant program, which is worth 
$348.0 million in FY 1982, requires that at least 90 percent of the funds be distributed to political 
subdivisions, nonprofit community organizations or migrant and seasonal farm worker organizations. 
Five percent is allowed for program transfer and five percent is allowed for administrative expenses. 


intended beneficiaries of block grant assistance, and will also create obligations 
for the direct recipient of such assistance, including the state, units of local 
government and private agencies.^' 

The Supreme Court's opinion in Pennhurst State School and Hospital v. 
Halderman^^ provides important guidance on when federal grant statutes — in- 
cluding those establishing block grants — are likely to be construed as creating 
"rights" or imposing "obligations" enforceable under section 1983. 

Pennhurst involved the interpretation of the Developmentally Disabled As- 
sistance and Bill of Rights Act.^^ The plaintiffs asserted that the Bill of Rights 
section of the Act — set forth in the form of legislative findings — required Penn- 
sylvania to establish individual treatment programs for each resident of Pennhurst 
State School and Hospital. The Supreme Court disagreed and concluded that the 
Act did not create rights nor impose obligations on a state that received federal 
grant assistance under the Act. 

The significance of Pennhurst to the new block grant programs can be seen 
from the rationale of Justice Rehnquist's majority opinion. Initially, Justice 
Rehnquist acknowledged the authority of Congress to establish grant programs 
and set conditions for the expenditure of funds under the spending power of the 
Constitution.^"^ But he reasoned that the implications of this authority are that: 

[LJegislation enacted pursuant to the spending power is much in the nature 
of contract: In return for federal funds, the states agree to comply with 
federally imposed conditions. The legitimacy of Congress' power to legislate 
under the spending power thus rests on whether the state voluntarily and 
knowingly accepts the terms of the 'contract. '^^ 

Justice Rehnquist continued: 

There can, of course, be no knowing acceptance if a state is unaware of 
the conditions or is unable to ascertain what is expected of it. Accordingly, 
if Congress intends to impose a condition on the grant of federal monies, 
it must do so unambiguously.^^ 

Justice Rehnquist concluded that this statement was a "canon" of statutory 
construction for all federal grant programs. 

5 1 . Federal courts have found third-party beneficiary contracts to exist in situations where a 
grantee promises or agrees to provide the benefits of grant assistance to another party. See, e.g., 
Lau V. Nichols, 414 U.S. 563, 568-69 (1974); Bossier Parish School Bd. v. Lemon, 370 F.2d 847 
(5th Cir.), cert, denied, 388 U.S. 911 (1967). 

52. 451 U.S. 1 (1981). 

53. See 42 U.S.C. §§ 6010(1) and (2) which provide that persons with developmental disa- 
bilities have a "right for appropriate treatment, services and rehabilitation for such disabilities . . . 
in a setting that is least restrictive of the person's liberty." 

54. The spending power is derived from Art. I, § 8, cl. 1 of the Constitution, which states 
that "Congress shall have the power to . . . provide for the . . . general welfare of the United 

55. 451 U.S. at 17. 

56. Id. at 17-18. 


The "knowing acceptance" principle of Pennhurst was recently cited by 
the U.S. Court of Appeals for the Third Circuit to support its decision not to 
give retroactive effect to section 185 of the Education Amendments of 1978^^ 
which authorizes the Department of Education to order repayment of federal 
funds misapplied or misspent by grant recipients. In State of New Jersey De- 
partment of Education v. Hufstedler,^^ the Department had contended that it 
could compel the repayment of grant funds under section 185, even though such 
funds had been awarded in 1976, five years prior to the enactment of section 
185. The Third Circuit acknowledged that Pennhurst was not directly applicable, 
but nevertheless concluded that: 

[T]he overarching principle of Pennhurst — that the terms and conditions 
of a federal grant must be set forth clearly and unambiguously in the statute 
enforcing the grant — precludes us from giving retroactive effect to a statute 
passed five years after the last disputed funds were received. ^^ 

At a time when there is vigorous opposition by the administration and grant 
recipients to excessive federal regulation of grant programs, Pennhurst provides 
a viable basis for challenging the broad range of conditions that now govern 
federal grant programs. However, Pennhurst also establishes a very important 
corollary: where a federal grant statute imposes unambiguous conditions — as 
many federal grant statutes do — state and local grantees will be legally obligated 
to perform such conditions. ^^ 

b. Remedies Under Section 1983 

National Sea Clammers has emerged as the pivotal decision for determining 
whether the ultimate beneficiaries of federal assistance will have remedies under 

57. 20 U.S. C. § 2835. 

58. 662 F.2d208 (3d Cir. 1981). 

59. Id. at p. 214. In contrast, just two days after the Third Circuit's decision in Hufstedler, 
the Fourth Circuit held that section 185 authorized the department to order West Virginia to repay 
federal funds determined to have been misspent by the state, even though the funds in question were 
received by the state in 1975. Notably, unlike the Third Circuit, the Fourth Circuit made no reference 
to the ''overarching principle of Pennhurst.'^ State of West Virginia v. Department of Education, 
667 F.2d417(4thCir. 1981). 

60. The Pennhurst decision continues to generate legal controversy despite the Court's 1981 
decision. Pennsylvania recently filed a petition for certiorari with the Supreme Court seeking review 
of the Third Circuit's refusal to order a refund of more than $1.2 million paid in fines by Pennsylvania. 
Pennhurst State School and Hospital v. Halderman, 673 F.2d 628 (3d Cir. 1982) (en banc), petition 
for cert, filed, 51 U.S.L.W. 3027 (U.S. June 18, 1982) (No. 81-2363). The fines were paid pursuant 
to a contempt citation issued against Pennsylvania for its failure to comply with a federal district 
court's order requiring Pennsylvania to fund Special Masters who were supervising the transfer of 
residents at Pennhurst State School and Hospital. See Halderman v. Pennhurst State School and 
Hospital, 533 F. Supp. 631 (E.D. Pa. 1981). Pennsylvania's failure to comply with the district 
court's order was based on the state legislature's refusal to grant a supplemental appropriation for 
$900,000 for special masters' salaries and expenses. The issue the Supreme Court is being asked 
to consider — whether the state's inability to comply with the district court's order because of the 
legislature's refusal to appropriate the necessary funds is a defense to the fines imposed on the state 
for civil contempt — raises a more fundamental question concerning the scope of the remedial powers 
of federal courts in actions brought against state officials under state law. 


section 1983. Under this case, third party beneficiaries can bring a section 1983 
action if they are denied benefits under a block grant, unless (1) Congress has 
provided an exclusive remedy, or (2) if the rights the beneficiary seeks to protect 
are not the kind of rights protectable under section 1983. 

A strong argument can be made that private suits under section 1983 may 
be brought to remedy violations of the statutory provisions of the 1981 block 
grant programs. A brief analysis of the Supreme Court's reasoning in National 
Sea Clammers supports this view. 

In National Sea Clammers, the Court indicated that legislative intent is the 
relevant inquiry under the exclusive remedy exception to section 1983 suits. ^' 
In that case, the Court concluded that the "elaborate enforcement provisions" 
under the Federal Water Pollution Control Act (FWPCA) and the Marine Pro- 
tection, Research and Sanctuaries Act (MPRSA) were evidence of congressional 

intent to ''supplant any remedy that otherwise would be available under section 
1983. "^2 

In the new block grants, as with most other grant programs. Congress has 
only provided limited administrative procedures for withholding funds from a 
state grantee that fails to meet the statutory requirements of a block grant. ^^ It 
is very unlikely that these remedies would be considered to constitute the "un- 
usually elaborate enforcement' ' mechanisms found to exist in National Sea Clam- 
mers.^ Furthermore, since the block grant statutes do not expressly authorize 
private suits, as was the case with the statutes at issue in National Sea Clammers, 
the withholding remedies alone would not appear to be the type of remedial 
devices "sufficiently comprehensive" to demonstrate congressional intent to 
preclude private suits under section 1983. 

61. 453 U.S. at 21. 

62. Id. Even before the Supreme Court's express disavowal of the availability of section 1983 
in situations where Congress has established an "exclusive remedy," lower courts relied upon the 
existence of such remedies to preclude section 1983 suits. See, e.g.. Uniformed Firefighters Ass'n 
V. City of New York, 512 F. Supp. 289 (S.D.N.Y. 1981); Meyerson v. Arizona, 501 F. Supp. 
859, 864 (D. Ariz. 1981). But see Smith v. City of Chicago, No. 79 C-3499 (N.D. 111. March 20, 

63. Congress has established a procedure whereby the Secretary of the Department of Health 
and Human Services is authorized to withhold grant funds from a state that fails to use such funds 
in accordance with the terms and conditions of the Community Services, Preventive Health, Primary 
Care, Maternal and Child Health, Low-Income Energy Assistance, and Alcohol, Drug Abuse and 
Mental Health Block Grants. See, e.g., OBRA supra note 41, §§ 679, 1907, 1917, and 1929. In 
addition, the Secretary of Housing and Urban Development is authorized to withhold funds used in 
violation of the terms and conditions of the Community Development Block Grant program. Id. § 

64. At least one expert on the subject of federal grant litigation has commented that the 
"exclusive remedy" exception may be inappropriate to apply to federal grant statutes in which 
typically the only remedies available are administrative sanctions such as suspension or termination 
of federal grant funds. In such cases, the individuals most directly affected — third party grant 
recipients — are afforded little participation in the administrative proceeding. Thus, it is unlikely that 
a court would find them "exclusive." See Brown, Pennhurst As A Source of Defenses for State and 
Local Governments, 31 Cath. U.L. Rev. 449 (1982). 


The second exception of National Sea Clammers raises the more difficult 
question of whether the 1981 block grants establish the kinds of "rights" en- 
forceable under section 1983. Neither Congress nor the Supreme Court has 
provided guidance on the nature of rights that may be protected under section 

Since section 1983 is a purely remedial statute and provides no substantive 
rights, ^^ a plaintiff in a block grant suit will have to show that the statute in 
question confers a right protectable under section 1983. Most grant statutes, 
including the block grant statutes, expressly identify the intended beneficiaries 
of federal assistance. ^^ Third parties would thus appear to have a strong argument 
that they do, in fact, have legal rights to the benefits of block grant assistance, 
and that such rights are enforceable under section 1983. 

Two recent district court opinions are particularly instructive on the range 
of potential judicial dispositions of section 1983 suits. These are BalfCo., Inc. 
V. Gaitor,^^ which demonstrates the degree to which a federal grant statute can 
be interpreted as creating legal rights, and Ryans v. New Jersey Comm'n For 
The Blind,^^ which involved the question of whether the availability of admin- 
istrative remedies under a grant program precluded a section 1983 suit based on 
the "exclusive remedy" test of National Sea Clammers. 

In Gaitor, a case involving the Federal- Aid Highway Act (FAHA),^^ the 
Balf Company sued the City of Hartford under section 1983, claiming that the 
city was required to satisfy federal regulations^' before implementing its decision 
to restrict vehicular traffic on a local access road financed primarily through 
FAHA funds and used extensively by Balf Company to transport its concrete 
mix and other stone products. The city defended on the grounds that Balf Com- 
pany lacked standing, arguing that based on Cort v. Ash,^'^ FAHA provided no 

65. The Court in National Sea Clammers did not reach the question of whether there existed 
rights protectable under section 1983, because it found that Congress had "foreclosed a § 1983 
remedy" by providing an exclusive remedy. 453 U.S. at 21. 

66. Chapman v. Houston Welfare Rights Organization, 441 U.S. 600. 617-18 (1979). 

67. See, e.g., the Alcohol, Drug Abuse and Mental Health Block Grant statute, which provides 
that in its annual application for assistance, a "state agrees to make grants to community mental 
health centers in the state for the provision of comprehensive mental health services — (A) principally 
to individuals . . . who are chronically ill . . ." and also "agrees to require that any community 
mental health center . . . provide — (A) outpatient services, including specialized services for chil- 
dren, the elderly, [and] individuals who are chronically mentally ill. . . ." OBRA supra note 41, 
§§ 1915(c)(3) and (4). See also Title I of the Rehabilitation Act of 1973, 29 U.S.C. § 723, which 
specifies a number of benefits and services which must be provided to eligible handicapped indi- 

68. 534 P. Supp. 600 (D. Conn. 1982). 

69. 542 F. Supp. 841 (D. N.J. 1982). 

70. 23 U.S.C. § 101 et seq. 

71. 23 C.F.R. § 620.203(a). 

72. 422 U.S. 66 (1975). In Cort v. Ash, the Supreme Court adopted a four factor test for 
analyzing whether a private cause of action can be implied under a particular statute. The first and 
most important of the Ash factors is whether the plaintiff is one of the class for whose especial 
benefit the statute was enacted. 


"right" upon which to base a section 1983 claim, since FAHA was not adopted 
for the "especial benefit" of the Balf Company. 

The district court characterized the issue in Gaitor as whether a "plaintiff 
in a section 1983 action must establish independent statutory rights and a cor- 
responding implied right to sue under that statute before such party can bring a 
section 1983 action, or alternatively, whether some lesser standard to confer 
standing is mandated after Maine v. Thiboutot/'^^ The court found that the Cort 
V. Ash analysis was inapplicable in a section 1983 action, reasoning: 

[I]f the plaintiff was unable to prove that the FAHA was adopted for its 
especial benefit, then presumably that plaintiff would be barred from suing 
the state defendants, even if the acts of those defendants violated federal 
law and seriously injured the plaintiff. Thiboutot and § 1983 thus would 
be rendered sui generis in these type of cases, a result that the Thiboutot 
Court by its language clearly did not contemplate.^"^ 

In light of this analysis, the court announced a separate test for recognizing 
rights enforceable under section 1983. The court held that a plaintiff in a section 
1983 action "need not establish that it has an implied right to sue under a separate 
federal statute, but rather must demonstrate that it has suffered an injury by the 
administration of a joint federal-state cooperative program and was an intended 
beneficiary of that program." ''^ (Emphasis in original.) 

The court went on to find that FAHA, though not adopted for the "especial 
benefit" of the Balf Company, "clearly was intended ... to benefit those 
persons who travel extensively in interstate commerce. "^^ Thus, the court ruled 
that Balf Company had standing to bring suit under section 1983, because it 
"may have been deprived of a 'right' secured by the laws of the United States, 
namely, the right to require [compliance] with FAHA regulations."^'' 

Gaitor is significant because it demonstrates the liberal degree to which a 
grant statute can be construed as creating a federally protectable right under 
section 1983. The decision in Ryans further extends the potential applicability 
of section 1983 to rights arising out of the administration of federal assistance 
programs by state and local officials because of its narrow reading of the "ex- 
clusive remedy" exception enunciated in National Sea Clammers. In Ryans a 
central issue was whether the administrative remedies of the Rehabilitation Act 
of 1973, 29 U.S.C. § 722(d), were exclusive and thus precluded a section 1983 
action to enforce rights created by Title I of the same Act.^^ 

73. 534 P. Supp. at 604. 

74. Id. 

75. Id. at 605. In its analysis, the court relied heavily on Yapalater v. Bates, 494 F. Supp. 
1349 (S.D.N.Y. 1980), where the court concluded that Thiboutot "necessarily limits the applicability 
of the Cort v. Ash analysis." Bates at 1357. The effect of Thiboutot being to "create a remedy 
where injury flows from a state's violation of governing federal law in a joint federal-state cooperative 
program." Bates at 1358. 

76. Id. 
11. Id. 

78. 542 F. Supp. at 844. 


The court turned to the RehabiHtation Act's legislative history in order to 
determine whether "a § 1983 remedy would be inconsistent with the statutory 
scheme" of 29 U.S.C. § 722(d) and whether the statute "suggests an intent on 
the part of Congress to foreclose" an action under section 1983. The court held 
that unlike the circumstances in National Sea Clammers — where the underlying 
statutes contained their own highly comprehensive set of enforcement remedies — 
"the procedures set forth in 29 U.S.C. § 722(d) and a civil action under § 1983 
would not be inconsistent but complementary."^^ In reaching this conclusion, 
the district court emphasized that it was "heeding" the "mandate" of Rosado 
V. Wyman,^^ where the Supreme Court held that courts should be: 

[M]ost reluctant to assume Congress has closed the avenue of effective 
judicial review to those individuals most directly affected by the adminis- 
tration of its program.^' 

Gaitor and Ryans are but two recent examples of a long line of cases in 
which third parties have sued state and local officials and agencies for denial of 
rights and benefits secured under a federal grant statute, ^^ as well as for failure 
to comply with the terms, conditions, and assurances relating to grant assis- 

Gaitor' s and Ryans' interpretation of National Sea Clammers have height- 
ened significance when one considers that most, if not all, federal grants establish 
literally scores of directives and conditions which state and local grant admin- 

79. Id. at 848. 29 U.S.C. § 722(d) requires the director of a state vocational rehabilitation 
agency to establish procedures for the review, upon the request of a handicapped individual, of 
determinations made by the state rehabilitation counselor respecting that individual's eligibility to 
receive benefits under Title I of the Rehabilitation Act. In addition, any handicapped person who 
is dissatisfied with the decision of the state agency director may request the Secretary of Health and 
Human Services (HHS) to review the state director's decision. In such cases, the Secretary is 
authorized to make recommendations to the state director as to the appropriate disposition of the 

80. 397 U.S. 397 (1970). 

81. Id. at 420. 

82. See generally Madden, Constitutional and Legal Foundations of Federal Grant Programs, 
Federal Grant Law, ABA (1981). See also Goldberg v. Kelly, 397 U.S. 254 (1970); South Eastern 
Human Development Corp. v. Schweiker, No. 82-1241 (8th Cir. Aug. 27, 1982); Montgomery 
Improvement Association v. HUD, 645 F.2d 291 (5th Cir. 1981); Lynch v. Maher, 507 F. Supp. 
1268 (D. Conn. 1981); Coalition for Block Grant Compliance v. Dept. of Housing and Urban 
Development, 450 F. Supp. 43 (E.D. Mich. 1978). 

83. See, e.g., Hendrick Hudson Central School District v. Rowley, 50U.S.L.W. 4925, (U.S. 
June 28, 1982), Lau v. Nichols, and Bossier Parish School Bd. v. Lemon, supra note 51; City of 
Englewood v. City of Los Angeles, 451 F.2d 948 (9th Cir. 1972). Apart from actions against state 
and local officials, it is well established that third parties may sue federal grantor agencies to assure 
that such agencies enforce the requirements of grant programs providing third party benefits or 
rights. See Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402 (1971); Adams v. Richardson, 
480 F.2d 1159 (D.C. Cir. 1973). But see Weems v. Pierce, 534 F. Supp. 740 (CD. 111. 1982). 
Finally, apart from suits brought by third parties, the federal government has ample authority to sue 
state and local grantees to enforce compliance with the conditions of grant assistance. See United 
States V. Marion County School District, 625 F.2d 607, reh. den., 629 F.2d 1350 (5th Cir. 1980). 


istrators must interpret and enforce. If these officials misinterpret or inadvertently 
disregard such directives and conditions, they are likely to subject themselves, 
or in the case of local officials their government employer, to suit under section 
1983 by an aggrieved third party. ^"^ The potential for such a result is perhaps 
most apparent in the case of the 1981 block grants. 

The Reagan Administration, in implementing its policy of giving state and 
local officials greater discretion to spend federal grant funds, has kept federal 
regulation of the 1981 block grants at a minimal level. ^^ However, without the 
guidance of federal regulations state and local officials will face difficult choices 
in interpreting and complying with the myriad of federal statutory directives and 
conditions accompanying block grant assistance. 

A recent decision by the U.S. Court of Appeals for the Eighth Circuit is 
an example of the confusion that can result over the interpretation of the obli- 
gations a federal grant statute may impose on a state that receives federal assis- 
tance. In South Eastern Human Development Corporation v. Schweiker,^^ the 
plaintiff, a nonprofit community services organization, sued South Dakota^^ for 
its failure to comply with the reporting and public hearing requirements of section 
1742 of OBRA. Title VI of OBRA established the Community Services Block 
Grant Program (CSBG). For fiscal year 1982 only, a state had the choice either 
to administer the CSBG program or to have its allotment distributed by HHS.^^ 
The Governor of South Dakota notified HHS of the state's intention to administer 
the CSBG program during FY 1982. However, no public hearings were held 
regarding FY 1982 Community Service funds and the state did not submit its 
report on the planned use of FY 1982 funds until December 11, 1981, more 
than two months after the commencement of FY 1982.^^ 

84. One congressional witness has testified that the decision in Maine v. Thiboutot gives rise 
too "double jeopardy" in the area of grants administration, since on the one hand, acceptance of 
a federal grant requires compliance with "many complex statutes and regulations," while on the 
other hand, "mandates from the judiciary" create the prospect of having to pay "millions of dollars 
to implement systematic relief." See Section 1983 Hearings, supra note 28, at 372 (statement of 
the Honorable J. Joseph Curran, representing the National Conference of State Legislatures). 

85. For example, the final regulations published by the Department of Health and Human 
Services covering seven of the nine new block grants emphasize flexibility and primary reliance on 
state discretion. See 47 Fed. Reg. 29472 (July 6, 1982). 

86. No. 82-1241 (8th Cir. Aug. 27, 1982). 

87. The suit named several defendants, including the Secretary of the U.S. Department of 
Health and Human Services, the Governor of South Dakota, and that state's planning bureau and 
its director. If the plaintiff had not named a federal defendant official, the case might well have 
been brought under section 1983. 

88. OBRA, supra note 41, § 682(b)(3). 

89. The dispute arose out of a series of decisions made by the Governor of South Dakota 
which culminated in the termination of grant funds for SEHDC. SEHDC claimed that because the 
reporting and public hearing requirements were not complied with, the Secretary of the Department 
of Health and Human Services (HHS) was obligated to distribute, in lieu of the state, the state's 
FY 1982 block grant allotment. Such action would have resulted in SEHDC's receiving two fiscal 
quarters' worth of community services funds. 


Rejecting the district court's conclusion that state grantees need not comply 
with section 1742,^^ the Eighth Circuit ruled that the most logical reading of 
section 1742 and its legislative history indicated that in adopting section 1742, 
Congress clearly intended to provide minimal protection for a state's citizens in 
the form of reporting and public participation mechanisms. Thus, the court 
ordered the Secretary of HHS to distribute to SEHDC the funds they would have 
received for the first two quarters of FY 1982. 

The problem of increased grant litigation — in which plaintiffs generally 
seek injunctive or other equitable relief^' — is only the tip of the section 1983 
iceberg: liability for money damages is now a foreseeable consequence for the 
mismanagement or misinterpretation of a federal grant program, since section 
1983 explicitly provides that "every person . . . shall be liable to the party 
injured." Moreover, several commentators have suggested that lower courts 
following Thiboutot and Owen may feel compelled to award damages to a pre- 
vailing party in a section 1983 grant suit.^^ 

In such suits, government officials will be insulated from liability for money 
damages to the extent that the circumstances surrounding the challenged action 
fall within Scheuef s analysis of the good-faith immunity defense. However, 
local governments would, as a result of Monell, Owen and Thiboutot, be liable 
for actual damages a third party may incur as a result of the mismanagement or 
improper administration of a program relating to a federal grant. ^^ 

Fortunately, for state and local officials, the threat of personal liability 
under section 1983 is offset in varying degrees by the existence at the state level 

90. Civ. 81-3072, (D.S.D. 1982). 

91. See, e.g., Lau v. Nichols, supra note 51; Lloyd v. Regional Transportation Authority, 
548 F.2d 1277 (7th Cir. 1977); Barrera v. Wheller, 475 F.2d 1338 (8th Cir. 1973), affd, 417 U.S. 
402 (1974); Fox v. Dept. of Housing and Urban Development, 468 F. Supp. 987 (E.D. Pa. 1980). 

92. See, e.g.. Section 1983 Hearings, supra note 28, at 95 (statement of Roy D. Bates on 
behalf of the National Institute of Municipal Law Officers); Id. at 336 (statement of Prof. George 
D. Brown). Professor Brown makes an interesting observation with respect to the types of situations 
in which damages might be awarded in grant suits. Professor Brown reasons that since the denial 
of a government benefit essential to a person's health or life itself could result in serious and 
demonstrable harm, such action could arguably be viewed as a common-law tort and thus properly 
subject to damages. It should be noted, however, that a successful argument along these lines requires 
first that there be a deprivation of some "right," as required by section 1983 itself. Whether a 
benefit conferred on an individual by a particular grant statute will rise to the level of a right protected 
by section 1983 will be a crucial, if not determinative, issue in all section 1983 grant disputes. This 
issue was analyzed supra in the text accompanying note 66. 

93 . Although comprehensive empirical data on the fiscal impact on municipalities of increased 
liability under sections 1983 and 1988 is not available, a 1981 survey by the National Institute of 
Municipal Law Officers (NIMLO) documents the potential liability of local governments for damages 
pursuant to section 1983. The NIMLO survey of more than 200 municipalities indicates that there 
is approximately $4.8 billion in pending civil rights claims against local governments. See Section 
1983 Hearings, supra note 28, at 120 (statement of Roy D. Bates on behalf of NIMLO). While 
this figure is staggering, the actual dollar amount of recoveries against municipalities sued under 
section 1983 is not known. 


of systems^"^ which provide (1) free defense counsel; (2) indemnification; (3) 
insurance; and (4) risk management training. ^^ For municipalities, however, the 
threat of section 1983 liability remains acute. ^^ 

c. Increased Litigation 

While the existence at the state level of systems designed to protect officials 
from personal liability may effectively serve in many instances to insulate the 
individual from financial disaster, other identifiable problems will persist. 

For one thing, the prospect of increased litigation is likely to dampen state 
or local government enthusiasm for taking innovative and adaptive approaches 
toward program management, as it would seem more likely than not that as 
officials exercise greater discretion, they increase the probability that their ad- 
ministration of a federally assisted program will be challenged. Program man- 
agers, when confronted with the prospect of increased liability, may introduce 
protective measures to insulate themselves from that risk. These measures, in 
the form of more elaborate procedures, extensive review and coordination mech- 
anisms, or manifold levels of decision and approval — all of which seem man- 
agerially proper — could add to existing levels of administrative inertia, delay 
and wasted resources. Consequently, government may become less effective as 
decision makers become more preoccupied with the possible consequences of a 
decision, rather than being concerned about the effect their decisions will have 
on the intended recipients of governmental assistance. 

A second effect of increased litigation under federal assistance programs 
relates to the degree to which delegation occurs in most government bureaucra- 
cies. The government official made a party to most section 1983 actions is often 
the highest elected or appointed official having responsibility over the function 
or person whose actions precipitated the alleged injury. In very large bureau- 
cracies, such as those servicing social assistance programs, the actions of sub- 
ordinate officials may be beyond the direct cognizance of the person in charge. 
As a result, higher ranking government officials will bear a disproportionate 
burden of liability solely because of their rank rather than because of any mis- 
conduct on their part. 

The third possible effect of increased litigation is a deterioration of state 
relations with the federal government. Presently, the federal government and its 

94. See infra text accompanying note 105. 

95. Risk management as used here, refers to a process in which sources of official liability 
are identified and analyzed in an effort to devise ways to avoid or reduce such liability. Risk 
management techniques can range from formal training programs and workshops to periodic seminars 
and informal question and answer sessions. Regardless of what form of risk management is practiced, 
its overriding benefit can be the reduction and elimination of the sources of official liability, resulting 
in a reduction in the costs of official misconduct to a state or local governmental employer. An 
overview discussion of risk management and its increasing popularity among municipalities is 
provided by Jaron, The Threat of Personal Liability Under the Federal Civil Rights Act: Does It 
Interfere with the Performance of State and Local Governments?, 13 Urb. Law. 1 (1981). 

96. For a review of the problems faced by municipalities, local officials and employees as a 
result of the failure (or unwillingness) of the insurance industry to provide adequate liability insurance 
coverage, see Jaron, supra note 95. 


officials retain much more immunity than state or local officials and munici- 
palities. This disparity in burden-sharing and liability between the federal "ac- 
tors" on one hand and state and local governments and officials on the other, 
can be a source of friction, and may result in an understandable reluctance by 
some state and local governments to assume responsibilities for federal assistance 
programs when the greatest legal exposure is directed toward them. 

A fourth consequence of increased litigation could be to heighten the already 
developing weariness of government officials who discover more and more of 
their time and energy being sapped in responding to legal claims and defending 
their actions under federal assistance programs, rather than in directing and 
overseeing tasks and subordinates. Consequently, state and local officials may 
choose not to administer federal assistance programs beyond what they are 
compelled to do.^^ 

Finally, it is widely acknowledged that the often mentioned fear of increased 
liability is no less chilling for public officials at the state and local levels than 
for those at the federal level. It could thereby discourage vigorous decision- 
making at lower levels of government. ^^ 

Giving state and local grantees and officials greater flexibility to manage 
federal programs is a broad policy decision supportable on many grounds. ^^ 
However, whatever policy changes are sought and implemented at the national 
level, '^^ it is clear that careful consideration should be given to the fact that 

97. Justice Powell, in his dissent in Thiboutot, perhaps best summarizes the negative impact 
that case could have on the future operation of federal grant and assistance programs: 

In practical effect of today's decision means that state and local governments, officers and 
employees now may face liability whenever a person believes he has been injured by the 
administration of any federal-state program, whether or not that program is related to equal 
or civil rights. 

448 U.S. at 22. The pernicious effect of Thiboutot has also been the topic of several scholarly 
articles. See, e.g., Peters, Municipal Liability After Owen v. City of Independence and Maine v. 
Thiboutot. 13 Urb. Law. 407 (1981); Groszyk & Madden, Managing Without Immunity: The Chal- 
lenge For State and Local Government Officials In the 1980s, Pub. Ad. Rev. (March/ April 1981). 

98. See Section 1983 Hearings, supra note 28, at 397 (statement of Scott Fosler and George 
Cross, representing the National Association of Counties). 

99. Executive Order 12372, issued July 14, 1982, is the most recent example of the current 
administration's policy of giving state and local governments more responsibility for administering 
federal financial assistance and development programs. E.O. 12372 revokes 0MB Circular A-95, 
relating to Evaluation, Review and Coordination of Federal and Federally Assisted Programs and 
Projects. The new order, to become effective April 30, 1983, strengthens the authority of state and 
local officials to establish federal assistance review procedures and priorities. 

100. Several proposals which relate to the federal grant system are awaiting congressional 
action. S. 807, the Federal Assistance Improvement Act of 1981, contains several provisions aimed 
at improving the federal grant system by simplifying federal assistance requirements to provide 
greater flexibility to state and local recipients. Although S. 807 has been before Congress in one 
form or another in recent years, and was recently reported by the Senate Committee on Government 
Affairs, it is unlikely that the bill will ever be acted upon so long as the President's New Federalism 
proposal remains a viable alternative. Another old idea has surfaced in the form of S. 10, which 


federal assistance programs frequently establish a myriad of rights, conditions 
and obligations which — according to current law — will be enforceable under 
section 1983. 

A relatively unexplored avenue for ameliorating the problem of increased 
litigation under section 1983 is discussed in the 1982 case of Patsy v. Board of 
Regents of State ofFlorida}^^ In Patsy, the Supreme Court held that a judicially- 
imposed requirement of exhaustion of state administrative remedies as a prereq- 
uisite to an action under section 1983 was improper. '^^ Importantly, however, 
in his opinion for the Court, Justice Marshall expressly acknowledged Congress' 
authority to require exhaustion of state administrative remedies as a prerequisite 
to bringing any action under section 1983,'^^ a view that was shared by three 
other Justices in their concurring opinions. '^"^ 

The Court's approval of such authority takes on added importance when 
considering viable solutions for avoiding increased section 1983 litigation in- 
volving federal assistance programs. 

For example, congressional enactment of an exhaustion requirement for 
claims arising out of state or local management of federal assistance programs 
would appear to offer a promising system for meeting the twin objectives of 
effective management of federal assistance programs and the vigorous enforce- 
ment of individual rights relating to such programs. 

passed the Senate last year and would establish a Commission for More Effective Government, 
patterned after the old Hoover Commissions. However, it is unlikely that the House Government 
Operations Committee will act on S. lO's companion bill — H.R. 18 — this session. In the event that 
the discussions on New Federalism reach a deadlock, or if Congress is unable to accept the legislation 
on the proposal, S. 807 and S. 10 may again become central topics of congressional debate. 

101. See note 6, supra. 

102. Such a requirement had been adopted below by the Fifth Circuit as well as other Courts 
of Appeals. See, e.g.. Patsy v. Florida International Univ., 634 F.2d 900 (5th Cir. 198 1) {en banc); 
Eisen v. Eastman, 421 F.2d 560 (2d Cir. 1969); Secret v. Brierton, 584 F.2d 823 (7th Cir. 1978). 

103. Justice Marshall relied in part on the legislative history of the Civil Rights of Institu- 
tionalized Persons Act, 42 U.S.C. § 1997(e), to hold that federal courts are generally precluded 
from requiring exhaustion of state administrative remedies under section 1983. Congress in enacting 
section 1997(e), according to Patsy, carved out a narrow exception to the general no-exhaustion 
rule to govern certain prisoner claims, and established a detailed scheme to ensure that state ad- 
ministrative remedies are adequate and effective. Justice Marshall concluded that a "judicially 
developed exhaustion rule in other [section 1983] cases" would be inconsistent with Congress' 
decision to adopt section 1997(e). 102 S. Ct. 2565. Justice Marshall observed that "Congress' 
superior institutional competence [suggests] that legislative not judicial solutions are preferable." 
Id. at 2567. He also emphasized that "it is not for us to say whether Congress . . . should create 
a similar scheme for other categories of § 1983 claims or whether Congress . . . should adopt an 
altogether different exhaustion requirement for nonprisoner § 1983 claims." Id. at 2567-68. 

104. Justice O'Connor, joined by Justice Rehnquist in her concurring opinion, noted that 
"perhaps Congress' enactment of [42 U.S.C. § 1997(e)], which creates a limited exhaustion re- 
quirement for prisoners bringing 8 1983 suits, will prompt it to reconsider the possibility of requiring 
exhaustion in the remainder of § 1983 cases." 102 S. Ct. at 2568. Similarly, Justice White, concurring 
in part, stated that "exhaustion is a statutory issue and the dispositive word on the matter belongs 
to Congress." Id. at 2569. 


Since section 1983 actions would not be precluded but merely postponed 
until administrative remedies have been exhausted, plaintiffs would retain the 
important right to have their claims heard by a federal court. Importantly, con- 
stitutional claims could be an exception to an exhaustion requirement, thus 
assuring no delay in the vindication of constitutional rights. 

State governments — in keeping with the trend giving them primary re- 
sponsibility for managing federal programs — would have a first-hand opportunity 
to correct whatever problems may be found to exist since the state's incentive 
to establish adequate administrative procedures will be strongest in situations 
where the state's potential for being sued under section 1983 is the greatest. 
Moreover, giving the states primary responsibility for handling complaints arising 
out of grant programs would substantially reduce the need to resort to federal 
courts for the resolution of disputes which often take years to resolve due to the 
inherent limitations of the adversary process. 

In addition, the existence of state administrative procedures is likely to 
heighten each state's awareness of the special needs of its citizenry at a time 
when states are being given more and more responsibility for administering 
federal assistance programs. The effective use of such procedures is likely over 
time to reduce the need for federal intervention, and at the same time allow the 
state to develop procedures which are best suited for handling complaints that 
arise within the state. 

Finally, federal courts could be authorized to waive the exhaustion require- 
ment in prescribed cases in order to ensure that state procedures made available 
are used to resolve those claims which are best suited for resolution in an 
administrative forum. 

B. State Responses to Section 1983 Liability 

The extensive experience at the state level in coping with section 1983 
liability provides an obviously worthwhile source of information for evaluating 
innovations that may have general applicability. Accordingly, a survey was 
conducted in which the fifty state attorneys general were asked to provide in- 
formation on existing systems directed toward ameliorating the problem of of- 
ficial liability. '^^ Specifically, each state attorney general was contacted to determine: 

( 1 ) Whether enterprise liability had been substituted for the personal liability 
of officials or employees in any respect? 

(2) Whether there exists a mechanism for either defending, indemnifying 
or insuring officials and employees in damage actions? 

(3) Whether there exists a risk management program designed to identify 
and reduce potential areas of liability exposure? 

(4) Whether any disciplinary procedures or administrative plans have been 
implemented to correct and reduce instances of official or employee 
misconduct that have or could result in liability? 

105. See Appendix B, which contains a copy of the survey letter sent to each attorney general. 
In many instances, telephone contacts provided useful supplements to the survey. 


(5) Whether, based on the state's activities and experience in the area of 
official immunity and liabiHty there are any suggestions or recommen- 
dations to be made with respect to the problem of liability of state or 
local officials for money damages? 

While the mechanisms in effect in any state are to some extent unique, the 
survey responses indicate several common ways in which the states seek to 
protect government officials from liability under section 1983. By way of sum- 
mary, many states now provide for the defense and indemnification of state 
officials and employees who acted in good faith. '^^ Without exception, however, 
the availability of these protections is contingent upon that official's having acted 
within the "course and scope of employment" and that his or her conduct was 
not ''willful and wanton," ''malicious," "malfeasant," or "grossly negligent." 
In effect, these state statutes follow closely the qualified immunity standard of 
Scheuer v. Rhodes. ^^^ In some states, the protective measures noted here extend 
to officials and employees of that state's political subdivisions. ^^^ In addition, 
many states authorize or require programs of insurance for state officials, em- 
ployees of state agencies, and persons employed by political subdivisions.'^^ 

No state appears to have embarked on a comprehensive shift toward reliance 
on enterprise liability to the extent reflected at the federal level in the proposed 
amendments to the FTC A. The survey reveals that state experience with risk 
management training programs, disciplinary proceedings, and similar approaches 
varies widely.''^ 

A description of the different state approaches to section 1983 liability and 
a summary discussion of the relevant statutory provisions in effect in each state 
is presented below.'" In cases where no survey response was received from a 
state's attorney general, the relevant provisions of that state code are briefly 
discussed, although without a survey response it was not possible to report the 
experience of such states under these provisions. 

106. See, e.g., Ariz. Rev. Stat. § 41-621, et seq.; Cal. Gov't. Code § 995; 10 Del. C. §§ 
3925 and 4001; Iowa C. § 25A.21; La. Rev. Stat. § 5108.1; Md. Code Ann. § 16C; Mass. Gen. 
Laws. Ann. ch. 258, §§ 2, 6, and 9; Nev. Rev. Stat. §§ 41.0339 and § 41.0349; N.J. Stat. Ann. 
§59.1-1 etseq.\Pa. Code §39.1 et seq.: Minn. Stat. §3.736; Wash. Rev. Code. Ann. §§4.92.060, 
4.92.070 and § 4.92.130; Wis. Stat. § 895.46. 

107. See Cass supra, at 1 172. 

108. See, e.g., Cal. Gov't. Code § 995; Fla. Stat. §§ 1 1 1.071 and 768.28(9)(a); Mass. Gen. 
Lawsch. 258, § 2; Nev. Rev. Stat. § 41.0339 and 41.0349; N.M. Stat. Ann. §41-4-1 etseq.;P2i. 
Code § 8521 et seq.; Wyo. Stat. § 1-39-101 et seq. 

109. See, e.g., Ga. Code Ann. §§ 89-973, 89-974 [O.C.G.A. §§ 45-9-1 through 45-9-5]; 
N.D. Cent. Code § 32-12.1-15. 

110. See, e.g.. Appendix B, Letters to Thomas J. Madden from Attorney Generals of the 
following states: Alaska, Kansas, Massachusetts, Mississippi, Missouri, Pennsylvania, South Car- 
olina, South Dakota, Texas, Vermont, Utah and Washington. But see Letter to Thomas J. Madden 
from Gerald L. Baliles, Attorney General, Commonwealth of Virginia. 

111. Analysis and comparison of state approaches is incomplete pending the expected im- 
minent receipt of additional responses. All of the responses received to date, along with supplementary 
materials, are contained in Appendix B. The thoroughness of some state survey responses has 
facilitated a more extended analysis of some states. 



93rd Congress 

H.R. 10439 Sponsored by Rodino, cosponsored by Hutchinson on 
September 20, 1973. To amend Title 28 U.S.C. to 
provide for exclusive remedy against the United 
States in suits based on acts or omissions of 
United States employees. 

September 20, 1973, referred to House 
Committee on the Judiciary. 

September 25, 1973, referred to Subcommittee on 
Claims and Governmental Relations. 

March 27, 1974, hearings before Subcommittee 

April 2, 1974, executive comment receive from 
the Unites States Postal Service. 

H.R. 12715 Sponsored by Chappell on February 7, 1974. To 
amend Title 28 U.S.C. to provide for exclusive 
remedy against the United States in suits based 
on acts or omissions of United States employees 

February 7, 1974, referred to House Committee 
on the Judiciary. 

February 11, 1974, referred to Subcommittee 
Claims and Governmental Relations. 

S. 2558 Sponsored by Hruska on October 10, 1973. To 

amend Title 28 U.S.C. to provide for exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 

October 10, 1973, referred to Senate Committee 
on the Judiciary. 

November 15, 1973, referred to Subcommittee 
Improvements in Judicial Machinery. 

April 9, 1974, executive comment received from 
the Treasury Department. 



95th Congress 

H.R. 9219 Sponsored by Rodino on September 20, 1977. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees . 

September 20, 1977, referred to House 
Committee on the Judiciary. 

September 29, 1977, referred to Subcommittee on 
Administrative Law and Governmental Relations. 

February 23, 1978, first day of Subcommittee 

May 17, 1978, final day of Subcommittee 

July 12, 1978, first day of Subcommittee 
consideration and markup. 

July 20, 1978, final day of Subcommittee 
consideration and markup. 

July 20, 1978, forwarded to full Committee, with 

H.R. 9437 Sponsored by Zeferetti on October 4, 1977. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States for suits based 
upon acts or omissions of United States 
officers, and employees routinely assigned to 
perform investigative, inspection or law 
enforcement functions. 

October 4, 1977, referred to House Committee 
on the Judiciary. 

October 7, 1977, referred to Subcommittee on 
Administrative Law and Governmental Relations. 

October 27, 1977, executive comment requested 
from the Agriculture Department, Labor 
Department, Treasury Department, FTC and FCC. 


- 3 - 

2117 Sponsored by Eastland on September 16, 1977. To 
amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 

September 16, 1977, referred to Senate 
Committee on the Judiciary. 

September 21, 1977, referred to Subcommittee 
on Citizens and Shareholders Rights. 

December 14, 1977, executive comment requested 
from the Justice Department. 

January 20, 1978, executive comment received 
from the Justice Department. 

January 26, 1978, first day of Subcommittee 

June 15, 1978, final day of Subcommittee 

S. 2868 Sponsored by Percy on April 10, 1978. To 

provide a remedy against the United States for 
torts arising under the Constitution or laws of 
the United States committed by officers or 
employees of the United States. 

April 10, 1978, referred to Senate Committee 
on the Judiciary. 

April 14, 1978, referred to Subcommittee on 
Citizens and Shareholders Rights. 

June 15, 1978, hearings held by Subcommittee. 

3314 Sponsored by Metzenbaum on July 18, 1978. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States employees 
and to amend Title 5 U.S.C. to permit a person 
aggrieved by a constitutional injury to initiate 
a disciplinary inquiry. 



- 4 - 

July 18, 1978, referred to Senate Committee on 
the Judiciary. 

July 19, 1978, referred to Subcommittee on 
Citizens and Shareholders Rights. 

July 26, 1978, Subcommittee reported measure 
without amendments to full Committee. 

96th Congress 

H.R. 193 

Sponsored by Chappell on January 15, 1 

1979. To amend Title 28 U.S.C. to provide for an 
exclusive remedy against the United States in 
suits based upon acts or commissions of United 
States employees. 

January 15, 1979, referred to House Committee 
on the Judiciary. 

February 27, 1979, ref-erred to Subcommittee on 
Administrative Law and Governmental 
Relations . 

June 20, 1979, for further action, see H.R. 



H.R. 2659 Sponsored by Rodino on March 3, 1979. To amend 
Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees, to provide a remedy against the United 
States with respect to constitutional torts and 
to establish procedures whereby a person injured 
by a constitutional tort may initiate and 
participate in a disciplinary inquiry with 
respect to such a tort. 

March 6, 1979, referred to House Committee on 
the Judiciary. 

March 14, 1979, referred to Subcommittee 
Administrative Law and Governmental 
Relations . 



- 5 - 

June 20, 1979, first day of Subcommittee 

August 1, 1979, final day of Subcommittee 

October 4, 1979, executive comment received from 
the Justice Department. 

October 31, 1979, Subcommittee consideration and 
markup session held, 

S. 695 Sponsored by Kennedy on March 15, 1979. To amend 
Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in actions based 
upon acts or omissions of United States 
employees, and to amend Title 5 U.S.C. to permit 
a person injured by a constitutional tort to 
initiate and participate in a disciplinary 
inquiry of the offending act or omission. 

March 15, 1979, referred to Senate Committee 
on the Judiciary. 

March 30, 1979, referred to Subcommittee on 
the Constitution. 

97th Congress 

H.R, 24 Sponsored by Danielson on January 5, 1981. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees, to provide a remedy against the United 
States with respect to constitutional torts, to 
establish procedures whereby a person injured by 
a constitutional tort may initiate and 
participate in a disciplinary inquiry with 
respect to such tort. 

January 5, 1981, referred to House Committee on 
the Judiciary. 

February 17, 1981, referred to Subcommittee on 
Administrative Law and Governmental Relations. 

February 25, 1981, executive comment requested 
from Justice Department, Treasury Department and 
Office of Personnel Management. 

October 13, 1981, subcommittee hearings held 
(Nov. 5, 1981; May 19, 20, 1982; May 26, 1982). 



- 6 - 

November 5, 1981, executive comment requested 
from Department of Defense. 

July 29, 1981, subcommittee consideration and 
markup session held. 

August 19, 1982, clean bill H.R. 7034 forwarded 
by subcommittee to full committee in lieu of 
H.R. 24. 

H.R. 1696 Sponsored by Chappell on February 5, 1981. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 

February 5, 1981, referred to House Committee on 
the Judiciary. 

February 17, 1981, referred to Subcommittee on 
Administrative Law and Governmental Relations. 

H.R. 6359 Sponsored by Kindness on May 12, 1982. A bill to 
amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees, to provide a remedy against the United 
States with respect to constitutional torts, and 
to establish procedures whereby a person injured 
by a constitutional tort may obtain a remedy. 

May 12, 1982, referred to House Committee on the 

May 17, 1982, referred to Subcommittee on 
Admnistrative Law and Governmental Relations. 

H.R. 7034 Sponsored by Hall on August 19, 1982. To amend 
Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees, to provide a remedy against the United 
States with respect to constitutional torts. 

August 19, 1982, referred to House Committee on 
the Judiciary. 

For previous action, see H.R. 24. 


- 7 - 

S. 1775 Sponsored by Grassley on October 26, 1981. To 

amend Title 28 U.S.C. to provide for an exclusive 
remedy against the United States in suits based 
upon acts or omissions of United States 
employees, to provide a remedy against the United 
States with respect to constitutional torts. 

October 26, 1981, referred to Senate Committee on 
the Judiciary. 

November 2, 1981, referred to Subcommittee on 
Agency Administration. 

November 13, 1981, subcommittee hearings held 
(November 16, 1981 and March 31, 1982). 

June 9, 1982, considered by Subcommittee on 
Agency Administration. 

September 23, 1983, subcommittee consideration 
and markup session held. 



Unlike many states, Alabama law does not provide for either the defense, 
indemnification or insurance of government officials and employees in suits 
arising out of an official's or employee's conduct that is within the scope of 


According to the Attorney General, Alaska has had very little litigation in 
which claims under section 1983 have been raised. However, the potential for 
official liability under section 1983 is viewed as a significant problem. Although 
Alaska, like most states, has not formally substituted enterprise liability for the 
personal liability of state officials, various collective bargaining agreements 
provide for state defense and indemnification of state officials and employees 
acting within the scope of their authority and responsibilities. 

The Attorney General reports that no special or formal disciplinary pro- 
cedures have been instituted; no formal provisions exist for the defense, indem- 
nification or insuring of state officials; and Alaska has not — except with respect 
to equal employment — instituted a risk management training program to avoid 
official or employee misconduct. 


Arizona has not substituted enterprise liability for the personal liability of 
officials or employees of the state. However, Arizona, pursuant to Ariz. Rev. 
Stat. § 41-621, et seq., provides insurance coverage for all officers, agents and 
employees against liability for acts or omissions of any nature while acting in 
the course and scope of employment or authorization. Additionally, Arizona law 
states that a state officer, agent or employee is not personally liable for an injury 
or damage resulting from his act or omission in a public official capacity where 
the exercise of discretion is involved and the official acted in good faith without 
wanton disregard of his statutory duties. Id. The Attorney General will defend 
a state official, agent or employee if self-insurance coverage is also provided. 

There is no state-wide training program to avoid potential liability under 
section 1983 actions, although the State's Risk Management Services relating 
to insurance does provide limited training through specialized seminars usually 
conducted by the Attorney General's office. 


The State of Arkansas will pay actual, but not punitive, damages adjudged 
or entered as a result of a compromise settlement, approved and recommended 
by the Attorney General, against officers or employees of the state based on an 
act or omission by the officer or employee while acting without malice and in 
good faith within the course and scope of employment and in the performance 
of his official duties. Ark. Stat. Ann. § 12-3401. 


California has not substituted enterprise liability for personal liability. Ac- 
cording to the Attorney General, the State of California is self-insured. In ad- 



dition, the state and all of its political subdivisions are required to provide for 
the defense of an employee or former employee in any civil action based on an 
act or omission arising out of the scope of his employment. Cal. Gov't Code 
§ 995. In such actions, either the state or the local governmental unit may refuse 
to defend the employee if he acted or failed to act because of fraud, corruption 
or malice, or if a conflict of interest would be created as a result of defending 
the employee. Id. § 995.2. 

In cases where the state or a local governmental unit defends the government 
employee, as well as those in which the employee conducts his own defense, 
the government is required to either pay or indemnify the employee for any 
judgment, settlement or compromise against him, once it is established that the 
injury arose out of an act or omission occurring in the scope of employment. 
Id. § 825 et seq. California law also authorizes local governmental units to insure 
any employee against all or any part of his liability resulting from an act or 
omission in the scope of his employment. Id. § 990. 

Informal risk management training exists, and the Attorney General keeps 
agencies informed of recent developments on liability and will, upon request, 
review and edit employee training manuals. 


Colorado has not substituted enterprise liability for the personal liability of 
state officers. The state's insurance policy, however, does provide coverage for 
actions against state officials and employees under 42 U.S.C. § 1983. 

In addition, a public entity — defined to include counties and cities — is 
required to indemnify an employee for any settlement or judgment against him 
in an action arising out of injuries due to an act or omission occurring during 
the performance of his duties and within the scope of employment, except where 
the employee's acts or omissions were willful or wanton. A public entity, how- 
ever, is not required to provide for the defense of its employees, but must bear 
the costs of a defense if the employee's actions occurred within the scope of 
employment and were not willful or wanton. Colo. Rev. Stat. § 24-10-110 et 

According to the Attorney General, the state has no risk management train- 
ing experience regarding official misconduct. 


Connecticut law deals with the liability of both state and municipal officers 
and employees. Conn. Gen. Stat. § 4-165 provides: 

No state officer or employee shall be personally liable for damage or injury, 
not wanton or wilful, caused in the performance of his duties and within 
the scope of his employment. Any person having a complaint for such 
damage or injury shall present it as a claim against the state under the 
provisions of this chapter. 

Similarly, section 7-465 makes local governments liable for damages caused by 
local officers or employees whose actions meet good faith standards of conduct. 



Like most states, Delaware has not substituted enterprise liability for the 
personal liability of state officials or employees. However, according to the 
Attorney General, Delaware law provides that public officers who act in con- 
nection with the determination of policy, in good faith and without gross or 
wanton negligence, are entitled to be indemnified for any judgment rendered 
against them and for the cost and expense of litigation, including attorney's fees. 
Del. Code Ann. tit. 10, § 4001 et seq. State officers and employees also have 
a right to representation. Id. § 3925. 

The Attorney General did not comment on Delaware's experience with risk 
management training for officials and employees. 

District of Columbia 

Although the District of Columbia does not provide for the defense or 
indemnification of its officers or employees, except in claims relating to an 
employee's negligent operation of a motor vehicle, D.C. Code Ann. § 1-1211 
et seq., the mayor is authorized to settle claims and suits against the District 
brought under section 1983 and arising out of the negligence or wrongful act of 
any officer or employee for whose acts the District would be liable if it were a 
private individual. See D.C. Code Ann. § 1-1202(1). 


Although Florida has not adopted enterprise liability as a substitute for 
official liability, officers and employees of the state and its political subdivisions 
are not personally liable in state court actions for their tortious acts within the 
course and scope of their employment so long as their actions do not exhibit 
bad faith, maliciousness or a wanton and willful disregard of human rights, 
safety or property. Fla. Stat. §§ 1 1 1.071 and 768.28(9)(a). Similarly, in federal 
civil rights suits under 42 U.S.C. § 1983, the state will pay any judgment entered 
against a state official unless it is determined that the official intentionally violated 
the plaintiff's rights. Id. §§ 1 1 1.071 and 284.38. The state also provides a legal 
defense for state officers and employees for such law suits arising out of their 
official duties. Id. ^ 11 1.07, while political subdivisions have the discretion, in 
certain cases, to provide for the representation of officials and employees. Id. 
§§ 111.06, 111.065. 

Any judgments against state officials and the cost of the defense of such 
actions are paid from the state's self-insurance trust fund. Id. §§ 1 1 1 .071 , 284.30, 
284.31. Since the Funds' inception in 1972, Florida has defended 2,524 section 
1983 damage actions against state officials at a cost of $2.5 million. The Attorney 
General points out, however, that this figure is somewhat misleading because it 
does not include the salary and support services for the assistant attorneys general 
who devote full time to section 1983 cases. 

The Attorney General did not comment on Florida's experience with risk 
management training. 


According to the Attorney General, the State of Georgia has established a 
self-insured indemnity program that provides coverage for state officials and 


employees. Ga. Code Ann. §§ 89-973, 89-974 [effective November 1, 1982, 
the State of Georgia is adopting a new code. The new cite will be O.C.G.A. 
§§ 45-9-1 through 45-9-5]. Local governments may also provide insurance cov- 
erage for local public officials and employees. Id. § 89-943. Notably, the Georgia 
State Merit System has a training division which offers courses on public per- 
sonnel management and instruction on public official and employee liability. 


The State of Hawaii has waived its immunity for liability for the torts of 
its employees and will assume liability to the same extent as a private individual 
under like circumstances. Hawaii Rev. Stat. § 662-2. However, § 662-2 is not 
applicable to "any claim based upon an act or omission of an employee of the 
state, exercising due care, in the execution of a statute or regulation." Thus, 
Hawaii officials sued in connection with the administration of a federal assistance 
program as interpreted by state law would appear to be subject to personal liability 
under section 1983. Hawaii Rev. Stat. §662-15. In most other cases, the Attorney 
General is authorized to defend any employee of the state in any civil action 
brought against the employee resulting from an act or omission while acting 
within the scope of his employment. Id. § 662-16. Finally, the State of Hawaii 
has waived its sovereign immunity and is covered by liability insurance in cases 
where section 662 is not applicable. Id. § 661-1 et seq. 


Idaho's Tort Claims Act, Chapter 9 of the Idaho Code, § 6-903 covers the 
liability of all governmental employees, including officers, employees and ser- 
vants of the state and its political subdivisions. In all civil actions, either the 
state or a political subdivision is liable, with certain exceptions, in lieu of the 

Idaho law requires the state to provide a comprehensive liability insurance 
plan to protect the state and its employees from claims and civil lawsuits. Id. § 
6-919. Political subdivisions are authorized to purchase the necessary liability 
insurance for themselves and their employees. Id. § 6-923. 


In 1981, Illinois enacted legislation to provide a program for purchasing 
liability insurance protection for state employees who act within the scope of 
their employment. 111. Ann. Stat., Ch. 127, § 6364. In addition, section 6364 
requires the State to defend, indemnify and hold harmless an employee against 
any claim in tort. Id. Chapter 85, § 2-302 of the 111. Ann. Stat, governs actions 
against employees of local governments. It provides: 

If any claim or action is instituted against an employee or former 
employee of a local public entity based on an injury allegedly arising out 
of an act or omission occurring within the scope of his employment as such 
employee, the entity may elect to do any one or more of the following: 

a) Appear and defend against the claim or action; 

b) Indemnify the employee or former employee for his court costs incurred 
in the defense of such claim or action; 


c) Pay, or indemnify the employee or former employee for a judgment 
based on such claim or action, or 

d) Pay, or indemnify the employee or former employee for, a compromise 
or settlement of such a claim or action. 


Chapter 16 of the Ind. Code Ann. covers the liability of government em- 
ployees of the State and its political subdivisions. With respect to indemnification 
and defense of employees, section 34-4-16.5-5 provides that the "governmental 
entity shall pay any judgment, compromise, or settlement of a claim or suit 
against an employee when the act or omission causing the loss is within the 
scope of his employment, regardless of whether the employee can or cannot be 
held personally liable for the loss and when the governor, in the case of a claim 
or suit against a state employee, or the governing body of the political subdi- 
vision, in the case of a claim or suits against an employee of political subdivision, 
determines that paying the judgment, compromise, or settlement is in the best 
interest of the governmental entity." 

Section 34-4-16.5-18 permits a governmental entity to purchase insurance 
to cover the liability of itself or its employees. Sections 34-4-16.7-1 and .7-2 
relate specifically to the personal civil liability under civil rights laws of em- 
ployees acting within the scope of employment and the defense of such employees 
for claims or suits arising under civil rights laws. 


The Attorney General reports that the most relevant Iowa statute bearing 
on the question of official liability is Iowa Code § 25A.21 (1981), part of the 
Iowa Tort Claims Act, which provides: 

The state shall defend and, except in cases of malfeasance in office or 
willful and wanton conduct, shall indemnify and hold harmless any em- 
ployee of the state against any claim as defined in section 25 A. 2, subsection 
5, paragraph 'b\ including claims arising under the Constitution, statutes, 
or rules of the United States or of any state. 

As with California and Florida, Iowa law also provides protections for local 
officials pursuant to the Municipal Tort Claims Act, Iowa Code § 613A.8 (1981). 
Under these statutes government attorneys defend state and local officials 
sued for damages pursuant to section 1983. The indemnity provisions, however, 
are basically limited to negligence claims and do not extend to claims grounded 
in "willful and wanton conduct." In addition, the indemnity provisions of Iowa 
law are largely inapplicable to section 1983 damage awards because such pro- 
visions are limited to negligence claims. No general insurance program is pro- 
vided or available to state employees for such risks, although the universities 
may have policies that would cover some such claims. But the Iowa State 
Association of Counties has developed a reinsurance program in which many 
counties participate that would cover many claims arising out of law enforcement 
activities. However, such coverage does not extend to the full range of section 
1983 exposure. 


Although that language has not been authoritatively construed by the Iowa 
Supreme Court, the Attorney General is of the view that section 1983 cases in 
which damages could be recovered against state officials would involve "willful 
and wanton conduct," and would thus preclude indemnification. 

The State of Iowa also self-insures against most risks and is developing a 
fairly comprehensive risk management program, but this program has not focused 
significantly as yet upon issues of personal liability for official misconduct. 

While there have been no major damage awards against state officials in 
Iowa, several pending cases involve damage claims against correctional officers 
at penal institutions for alleged assaults against inmates. Were those to be suc- 
cessful, the officers would be personally liable for the damages. 

Unlike most states, Iowa law establis